llvm/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-llvm.amdgcn.image.sample.a16.ll

; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -stop-after=legalizer -o - %s | FileCheck -check-prefix=GFX9 %s
; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 -stop-after=legalizer -o - %s | FileCheck -check-prefix=GFX10 %s
; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -stop-after=legalizer -o - %s | FileCheck -check-prefix=GFX11 %s
; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1200 -stop-after=legalizer -o - %s | FileCheck -check-prefix=GFX12 %s

define amdgpu_ps <4 x float> @sample_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s) {
  ; GFX9-LABEL: name: sample_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f16(i32 15, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t) {
  ; GFX9-LABEL: name: sample_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.2d.v4f32.f16(i32 15, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %r) {
  ; GFX9-LABEL: name: sample_3d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_3d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_3d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_3d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.3d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.3d.v4f32.f16(i32 15, half %s, half %t, half %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_cube(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %face) {
  ; GFX9-LABEL: name: sample_cube
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_cube
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_cube
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_cube
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cube), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.cube.v4f32.f16(i32 15, half %s, half %t, half %face, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_1darray(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %slice) {
  ; GFX9-LABEL: name: sample_1darray
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1darray), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_1darray
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1darray), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_1darray
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1darray), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_1darray
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1darray), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.1darray.v4f32.f16(i32 15, half %s, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_2darray(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %slice) {
  ; GFX9-LABEL: name: sample_2darray
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_2darray
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_2darray
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_2darray
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2darray), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.2darray.v4f32.f16(i32 15, half %s, half %t, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s) {
  ; GFX9-LABEL: name: sample_c_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.1d.v4f32.f16(i32 15, float %zcompare, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t) {
  ; GFX9-LABEL: name: sample_c_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %clamp) {
  ; GFX9-LABEL: name: sample_cl_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_cl_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_cl_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_cl_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.cl.1d.v4f32.f16(i32 15, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %clamp) {
  ; GFX9-LABEL: name: sample_cl_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_cl_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_cl_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_cl_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.cl.2d.v4f32.f16(i32 15, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %clamp) {
  ; GFX9-LABEL: name: sample_c_cl_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_cl_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_cl_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_cl_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.cl.1d.v4f32.f16(i32 15, float %zcompare, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t, half %clamp) {
  ; GFX9-LABEL: name: sample_c_cl_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_cl_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_cl_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_cl_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.cl.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_b_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s) {
  ; GFX9-LABEL: name: sample_b_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_b_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_b_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_b_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.b.1d.v4f32.f16.f16(i32 15, half %bias, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_b_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s, half %t) {
  ; GFX9-LABEL: name: sample_b_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_b_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_b_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_b_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.b.2d.v4f32.f16.f16(i32 15, half %bias, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_b_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s) {
  ; GFX9-LABEL: name: sample_c_b_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_b_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_b_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_b_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.b.1d.v4f32.f16.f16(i32 15, half %bias, float %zcompare, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_b_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s, half %t) {
  ; GFX9-LABEL: name: sample_c_b_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_b_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_b_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_b_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.b.2d.v4f32.f16.f16(i32 15, half %bias, float %zcompare, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_b_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s, half %clamp) {
  ; GFX9-LABEL: name: sample_b_cl_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_b_cl_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_b_cl_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_b_cl_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.b.cl.1d.v4f32.f16.f16(i32 15, half %bias, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_b_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s, half %t, half %clamp) {
  ; GFX9-LABEL: name: sample_b_cl_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_b_cl_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_b_cl_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_b_cl_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.b.cl.2d.v4f32.f16.f16(i32 15, half %bias, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_b_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s, half %clamp) {
  ; GFX9-LABEL: name: sample_c_b_cl_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_b_cl_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_b_cl_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_b_cl_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.b.cl.1d.v4f32.f16.f16(i32 15, half %bias, float %zcompare, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_b_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s, half %t, half %clamp) {
  ; GFX9-LABEL: name: sample_c_b_cl_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_b_cl_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_b_cl_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_b_cl_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.b.cl.2d.v4f32.f16.f16(i32 15, half %bias, float %zcompare, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s) {
  ; GFX9-LABEL: name: sample_d_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_d_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_d_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_d_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
  ; GFX9-LABEL: name: sample_d_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX9-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_d_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX10-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_d_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX11-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_d_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX12-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.d.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_d_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, half %s, half %t, half %r) {
  ; GFX9-LABEL: name: sample_d_3d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX9-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX9-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX9-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX9-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX9-NEXT:   [[TRUNC7:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX9-NEXT:   [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
  ; GFX9-NEXT:   [[TRUNC8:%[0-9]+]]:_(s16) = G_TRUNC [[COPY20]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_d_3d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX10-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX10-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX10-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX10-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX10-NEXT:   [[TRUNC7:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX10-NEXT:   [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
  ; GFX10-NEXT:   [[TRUNC8:%[0-9]+]]:_(s16) = G_TRUNC [[COPY20]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_d_3d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX11-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX11-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX11-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX11-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX11-NEXT:   [[TRUNC7:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX11-NEXT:   [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
  ; GFX11-NEXT:   [[TRUNC8:%[0-9]+]]:_(s16) = G_TRUNC [[COPY20]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_d_3d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX12-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX12-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX12-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX12-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX12-NEXT:   [[TRUNC7:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX12-NEXT:   [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
  ; GFX12-NEXT:   [[TRUNC8:%[0-9]+]]:_(s16) = G_TRUNC [[COPY20]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR5]](<2 x s16>), [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, half %s, half %t, half %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s) {
  ; GFX9-LABEL: name: sample_c_d_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_d_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_d_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_d_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
  ; GFX9-LABEL: name: sample_c_d_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX9-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX9-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX9-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_d_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX10-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX10-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX10-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_d_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX11-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX11-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX11-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_d_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX12-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX12-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX12-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s, half %clamp) {
  ; GFX9-LABEL: name: sample_d_cl_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_d_cl_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_d_cl_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_d_cl_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.d.cl.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
  ; GFX9-LABEL: name: sample_d_cl_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX9-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX9-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX9-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_d_cl_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX10-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX10-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX10-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_d_cl_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX11-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX11-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX11-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_d_cl_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX12-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX12-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX12-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.d.cl.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp) {
  ; GFX9-LABEL: name: sample_c_d_cl_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_d_cl_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_d_cl_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_d_cl_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
  ; GFX9-LABEL: name: sample_c_d_cl_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX9-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX9-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX9-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX9-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX9-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<10 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[CONCAT_VECTORS]](<10 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_d_cl_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX10-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX10-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX10-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX10-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX10-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_d_cl_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX11-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX11-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX11-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX11-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX11-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_d_cl_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX12-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX12-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX12-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX12-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX12-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s) {
  ; GFX9-LABEL: name: sample_cd_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_cd_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_cd_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_cd_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.cd.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
  ; GFX9-LABEL: name: sample_cd_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX9-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_cd_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX10-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_cd_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX11-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_cd_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX12-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.cd.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s) {
  ; GFX9-LABEL: name: sample_c_cd_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_cd_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_cd_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_cd_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
  ; GFX9-LABEL: name: sample_c_cd_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX9-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX9-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX9-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_cd_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX10-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX10-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX10-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_cd_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX11-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX11-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX11-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_cd_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX12-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX12-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX12-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s, half %clamp) {
  ; GFX9-LABEL: name: sample_cd_cl_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_cd_cl_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_cd_cl_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_cd_cl_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.cd.cl.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
  ; GFX9-LABEL: name: sample_cd_cl_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX9-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX9-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX9-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_cd_cl_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX10-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX10-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX10-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_cd_cl_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX11-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX11-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX11-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_cd_cl_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX12-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX12-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX12-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.cd.cl.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp) {
  ; GFX9-LABEL: name: sample_c_cd_cl_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_cd_cl_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_cd_cl_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_cd_cl_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
  ; GFX9-LABEL: name: sample_c_cd_cl_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX9-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX9-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX9-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX9-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX9-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<10 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[CONCAT_VECTORS]](<10 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_cd_cl_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX10-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX10-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX10-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX10-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX10-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_cd_cl_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX11-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX11-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX11-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX11-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX11-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_cd_cl_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX12-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX12-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX12-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX12-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX12-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_l_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %lod) {
  ; GFX9-LABEL: name: sample_l_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_l_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_l_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_l_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.l.1d.v4f32.f16(i32 15, half %s, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_l_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %lod) {
  ; GFX9-LABEL: name: sample_l_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_l_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_l_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_l_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.l.2d.v4f32.f16(i32 15, half %s, half %t, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_l_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %lod) {
  ; GFX9-LABEL: name: sample_c_l_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_l_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_l_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_l_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.l.1d.v4f32.f16(i32 15, float %zcompare, half %s, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_l_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t, half %lod) {
  ; GFX9-LABEL: name: sample_c_l_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_l_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_l_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_l_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.l.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_lz_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s) {
  ; GFX9-LABEL: name: sample_lz_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_lz_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_lz_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_lz_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.lz.1d.v4f32.f16(i32 15, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_lz_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t) {
  ; GFX9-LABEL: name: sample_lz_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_lz_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_lz_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_lz_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.lz.2d.v4f32.f16(i32 15, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_lz_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s) {
  ; GFX9-LABEL: name: sample_c_lz_1d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_lz_1d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_lz_1d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_lz_1d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.lz.1d.v4f32.f16(i32 15, float %zcompare, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps <4 x float> @sample_c_lz_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t) {
  ; GFX9-LABEL: name: sample_c_lz_2d
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX9-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX10-LABEL: name: sample_c_lz_2d
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX10-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX11-LABEL: name: sample_c_lz_2d
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX11-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
  ;
  ; GFX12-LABEL: name: sample_c_lz_2d
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   $vgpr2 = COPY [[UV2]](s32)
  ; GFX12-NEXT:   $vgpr3 = COPY [[UV3]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body:
  %v = call <4 x float> @llvm.amdgcn.image.sample.c.lz.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <4 x float> %v
}

define amdgpu_ps float @sample_c_d_o_2darray_V1(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice) {
  ; GFX9-LABEL: name: sample_c_d_o_2darray_V1
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX9-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX9-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX9-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX9-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX9-NEXT:   [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
  ; GFX9-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY20]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (s32), addrspace 8)
  ; GFX9-NEXT:   $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0
  ;
  ; GFX10-LABEL: name: sample_c_d_o_2darray_V1
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX10-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX10-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX10-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX10-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX10-NEXT:   [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
  ; GFX10-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY20]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (s32), addrspace 8)
  ; GFX10-NEXT:   $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0
  ;
  ; GFX11-LABEL: name: sample_c_d_o_2darray_V1
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX11-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX11-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX11-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX11-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX11-NEXT:   [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
  ; GFX11-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY20]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (s32), addrspace 8)
  ; GFX11-NEXT:   $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0
  ;
  ; GFX12-LABEL: name: sample_c_d_o_2darray_V1
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX12-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX12-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX12-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX12-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX12-NEXT:   [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
  ; GFX12-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY20]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (s32), addrspace 8)
  ; GFX12-NEXT:   $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0
main_body:
  %v = call float @llvm.amdgcn.image.sample.c.d.o.2darray.f32.f16.f16(i32 4, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret float %v
}

define amdgpu_ps <2 x float> @sample_c_d_o_2darray_V2(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice) {
  ; GFX9-LABEL: name: sample_c_d_o_2darray_V2
  ; GFX9: bb.1.main_body:
  ; GFX9-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
  ; GFX9-NEXT: {{  $}}
  ; GFX9-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX9-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX9-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX9-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX9-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX9-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX9-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX9-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX9-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX9-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX9-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX9-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX9-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX9-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX9-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX9-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX9-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX9-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX9-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX9-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX9-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX9-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX9-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX9-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX9-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX9-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX9-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX9-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX9-NEXT:   [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
  ; GFX9-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY20]](s32)
  ; GFX9-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX9-NEXT:   [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX9-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX9-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX9-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX9-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX9-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX9-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<2 x s32>), addrspace 8)
  ; GFX9-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
  ; GFX9-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX9-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
  ;
  ; GFX10-LABEL: name: sample_c_d_o_2darray_V2
  ; GFX10: bb.1.main_body:
  ; GFX10-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
  ; GFX10-NEXT: {{  $}}
  ; GFX10-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX10-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX10-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX10-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX10-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX10-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX10-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX10-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX10-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX10-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX10-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX10-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX10-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX10-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX10-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX10-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX10-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX10-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX10-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX10-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX10-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX10-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX10-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX10-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX10-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX10-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX10-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX10-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX10-NEXT:   [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
  ; GFX10-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY20]](s32)
  ; GFX10-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX10-NEXT:   [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX10-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX10-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX10-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX10-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX10-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX10-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<2 x s32>), addrspace 8)
  ; GFX10-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
  ; GFX10-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX10-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX10-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
  ;
  ; GFX11-LABEL: name: sample_c_d_o_2darray_V2
  ; GFX11: bb.1.main_body:
  ; GFX11-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
  ; GFX11-NEXT: {{  $}}
  ; GFX11-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX11-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX11-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX11-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX11-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX11-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX11-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX11-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX11-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX11-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX11-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX11-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX11-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX11-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX11-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX11-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX11-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX11-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX11-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX11-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX11-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX11-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX11-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX11-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX11-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX11-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX11-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX11-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX11-NEXT:   [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
  ; GFX11-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY20]](s32)
  ; GFX11-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX11-NEXT:   [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX11-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX11-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX11-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX11-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX11-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX11-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<2 x s32>), addrspace 8)
  ; GFX11-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
  ; GFX11-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX11-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX11-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
  ;
  ; GFX12-LABEL: name: sample_c_d_o_2darray_V2
  ; GFX12: bb.1.main_body:
  ; GFX12-NEXT:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
  ; GFX12-NEXT: {{  $}}
  ; GFX12-NEXT:   [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
  ; GFX12-NEXT:   [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
  ; GFX12-NEXT:   [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
  ; GFX12-NEXT:   [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
  ; GFX12-NEXT:   [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
  ; GFX12-NEXT:   [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
  ; GFX12-NEXT:   [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
  ; GFX12-NEXT:   [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
  ; GFX12-NEXT:   [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
  ; GFX12-NEXT:   [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
  ; GFX12-NEXT:   [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
  ; GFX12-NEXT:   [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
  ; GFX12-NEXT:   [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
  ; GFX12-NEXT:   [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
  ; GFX12-NEXT:   [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
  ; GFX12-NEXT:   [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
  ; GFX12-NEXT:   [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
  ; GFX12-NEXT:   [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
  ; GFX12-NEXT:   [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
  ; GFX12-NEXT:   [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
  ; GFX12-NEXT:   [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
  ; GFX12-NEXT:   [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
  ; GFX12-NEXT:   [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
  ; GFX12-NEXT:   [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
  ; GFX12-NEXT:   [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
  ; GFX12-NEXT:   [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
  ; GFX12-NEXT:   [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
  ; GFX12-NEXT:   [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
  ; GFX12-NEXT:   [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
  ; GFX12-NEXT:   [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY20]](s32)
  ; GFX12-NEXT:   [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
  ; GFX12-NEXT:   [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
  ; GFX12-NEXT:   [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
  ; GFX12-NEXT:   [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
  ; GFX12-NEXT:   [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
  ; GFX12-NEXT:   [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
  ; GFX12-NEXT:   [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
  ; GFX12-NEXT:   [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<2 x s32>), addrspace 8)
  ; GFX12-NEXT:   [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
  ; GFX12-NEXT:   $vgpr0 = COPY [[UV]](s32)
  ; GFX12-NEXT:   $vgpr1 = COPY [[UV1]](s32)
  ; GFX12-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
main_body:
  %v = call <2 x float> @llvm.amdgcn.image.sample.c.d.o.2darray.v2f32.f32.f16(i32 6, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
  ret <2 x float> %v
}

declare <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f16(i32, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <8 x float> @llvm.amdgcn.image.sample.1d.v8f32.f16(i32, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.2d.v4f32.f16(i32, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.3d.v4f32.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.cube.v4f32.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.1darray.v4f32.f16(i32, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.2darray.v4f32.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1

declare <4 x float> @llvm.amdgcn.image.sample.c.1d.v4f32.f16(i32, float, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.2d.v4f32.f16(i32, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.cl.1d.v4f32.f16(i32, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.cl.2d.v4f32.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.cl.1d.v4f32.f16(i32, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.cl.2d.v4f32.f16(i32, float, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1

declare <4 x float> @llvm.amdgcn.image.sample.b.1d.v4f32.f16.f16(i32, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.b.2d.v4f32.f16.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.b.1d.v4f32.f16.f16(i32, half, float, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.b.2d.v4f32.f16.f16(i32, half, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.b.cl.1d.v4f32.f16.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.b.cl.2d.v4f32.f16.f16(i32, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.b.cl.1d.v4f32.f16.f16(i32, half, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.b.cl.2d.v4f32.f16.f16(i32, half, float, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1

declare <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.d.2d.v4f32.f16.f16(i32, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f16.f16(i32, half, half, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.d.1d.v4f32.f32.f16(i32, float, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.d.2d.v4f32.f32.f16(i32, float, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.d.cl.1d.v4f32.f16.f16(i32, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.d.cl.2d.v4f32.f16.f16(i32, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.d.cl.1d.v4f32.f32.f16(i32, float, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.d.cl.2d.v4f32.f32.f16(i32, float, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1

declare <4 x float> @llvm.amdgcn.image.sample.cd.1d.v4f32.f16.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.cd.2d.v4f32.f16.f16(i32, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.cd.1d.v4f32.f32.f16(i32, float, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.cd.2d.v4f32.f32.f16(i32, float, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.cd.cl.1d.v4f32.f16.f16(i32, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.cd.cl.2d.v4f32.f16.f16(i32, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.1d.v4f32.f32.f16(i32, float, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.2d.v4f32.f32.f16(i32, float, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1

declare <4 x float> @llvm.amdgcn.image.sample.l.1d.v4f32.f16(i32, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.l.2d.v4f32.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.l.1d.v4f32.f16(i32, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.l.2d.v4f32.f16(i32, float, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1

declare <4 x float> @llvm.amdgcn.image.sample.lz.1d.v4f32.f16(i32, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.lz.2d.v4f32.f16(i32, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.lz.1d.v4f32.f16(i32, float, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <4 x float> @llvm.amdgcn.image.sample.c.lz.2d.v4f32.f16(i32, float, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1

declare float @llvm.amdgcn.image.sample.c.d.o.2darray.f32.f16.f16(i32, i32, float, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1
declare <2 x float> @llvm.amdgcn.image.sample.c.d.o.2darray.v2f32.f32.f16(i32, i32, float, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) #1

attributes #0 = { nounwind }
attributes #1 = { nounwind readonly }
attributes #2 = { nounwind readnone }