llvm/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-fcmp.mir

# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
# RUN: llc -O0 -mtriple=amdgcn -mcpu=hawaii  -run-pass=legalizer -verify-machineinstrs -o - %s | FileCheck -check-prefix=GFX7 %s
# RUN: llc -O0 -mtriple=amdgcn -mcpu=fiji  -run-pass=legalizer -verify-machineinstrs -o - %s | FileCheck -check-prefix=GFX8 %s
# RUN: llc -O0 -mtriple=amdgcn -mcpu=gfx900  -run-pass=legalizer -verify-machineinstrs -o - %s | FileCheck -check-prefix=GFX9 %s
# RUN: llc -O0 -mtriple=amdgcn -mcpu=gfx1010  -run-pass=legalizer -verify-machineinstrs -o - %s | FileCheck -check-prefix=GFX9 %s
# RUN: llc -O0 -mtriple=amdgcn -mcpu=gfx1100  -run-pass=legalizer -verify-machineinstrs -o - %s | FileCheck -check-prefix=GFX9 %s
# RUN: llc -O0 -mtriple=amdgcn -mcpu=gfx1150  -run-pass=legalizer -verify-machineinstrs -o - %s | FileCheck -check-prefix=GFX9 %s

---
name: test_fcmp_s32
body: |
  bb.0:
    liveins: $vgpr0
    ; GFX7-LABEL: name: test_fcmp_s32
    ; GFX7: liveins: $vgpr0
    ; GFX7-NEXT: {{  $}}
    ; GFX7-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
    ; GFX7-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
    ; GFX7-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s32), [[COPY]]
    ; GFX7-NEXT: [[SELECT:%[0-9]+]]:_(s32) = G_SELECT [[FCMP]](s1), [[C]], [[COPY]]
    ; GFX7-NEXT: $vgpr0 = COPY [[SELECT]](s32)
    ; GFX8-LABEL: name: test_fcmp_s32
    ; GFX8: liveins: $vgpr0
    ; GFX8-NEXT: {{  $}}
    ; GFX8-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
    ; GFX8-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
    ; GFX8-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s32), [[COPY]]
    ; GFX8-NEXT: [[SELECT:%[0-9]+]]:_(s32) = G_SELECT [[FCMP]](s1), [[C]], [[COPY]]
    ; GFX8-NEXT: $vgpr0 = COPY [[SELECT]](s32)
    ; GFX9-LABEL: name: test_fcmp_s32
    ; GFX9: liveins: $vgpr0
    ; GFX9-NEXT: {{  $}}
    ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
    ; GFX9-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
    ; GFX9-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s32), [[COPY]]
    ; GFX9-NEXT: [[SELECT:%[0-9]+]]:_(s32) = G_SELECT [[FCMP]](s1), [[C]], [[COPY]]
    ; GFX9-NEXT: $vgpr0 = COPY [[SELECT]](s32)
    %0:_(s32) = G_CONSTANT i32 0
    %1:_(s32) = COPY $vgpr0
    %2:_(s1) = G_FCMP floatpred(oeq), %0, %1
    %3:_(s32) = G_SELECT %2, %0, %1
    $vgpr0 = COPY %3
...

---
name: test_fcmp_s64
body: |
  bb.0:
    liveins: $vgpr0_vgpr1
    ; GFX7-LABEL: name: test_fcmp_s64
    ; GFX7: liveins: $vgpr0_vgpr1
    ; GFX7-NEXT: {{  $}}
    ; GFX7-NEXT: [[C:%[0-9]+]]:_(s64) = G_CONSTANT i64 0
    ; GFX7-NEXT: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
    ; GFX7-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s64), [[COPY]]
    ; GFX7-NEXT: [[SELECT:%[0-9]+]]:_(s64) = G_SELECT [[FCMP]](s1), [[C]], [[COPY]]
    ; GFX7-NEXT: $vgpr0_vgpr1 = COPY [[SELECT]](s64)
    ; GFX8-LABEL: name: test_fcmp_s64
    ; GFX8: liveins: $vgpr0_vgpr1
    ; GFX8-NEXT: {{  $}}
    ; GFX8-NEXT: [[C:%[0-9]+]]:_(s64) = G_CONSTANT i64 0
    ; GFX8-NEXT: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
    ; GFX8-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s64), [[COPY]]
    ; GFX8-NEXT: [[SELECT:%[0-9]+]]:_(s64) = G_SELECT [[FCMP]](s1), [[C]], [[COPY]]
    ; GFX8-NEXT: $vgpr0_vgpr1 = COPY [[SELECT]](s64)
    ; GFX9-LABEL: name: test_fcmp_s64
    ; GFX9: liveins: $vgpr0_vgpr1
    ; GFX9-NEXT: {{  $}}
    ; GFX9-NEXT: [[C:%[0-9]+]]:_(s64) = G_CONSTANT i64 0
    ; GFX9-NEXT: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
    ; GFX9-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s64), [[COPY]]
    ; GFX9-NEXT: [[SELECT:%[0-9]+]]:_(s64) = G_SELECT [[FCMP]](s1), [[C]], [[COPY]]
    ; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[SELECT]](s64)
    %0:_(s64) = G_CONSTANT i64 0
    %1:_(s64) = COPY $vgpr0_vgpr1
    %2:_(s1) = G_FCMP floatpred(oeq), %0, %1
    %3:_(s64) = G_SELECT %2, %0, %1
    $vgpr0_vgpr1 = COPY %3
...

---
name: test_fcmp_s16
body: |
  bb.0:
    liveins: $vgpr0
    ; GFX7-LABEL: name: test_fcmp_s16
    ; GFX7: liveins: $vgpr0
    ; GFX7-NEXT: {{  $}}
    ; GFX7-NEXT: [[C:%[0-9]+]]:_(s16) = G_CONSTANT i16 0
    ; GFX7-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
    ; GFX7-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY]](s32)
    ; GFX7-NEXT: [[FPEXT:%[0-9]+]]:_(s32) = G_FPEXT [[C]](s16)
    ; GFX7-NEXT: [[FPEXT1:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC]](s16)
    ; GFX7-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[FPEXT]](s32), [[FPEXT1]]
    ; GFX7-NEXT: [[SELECT:%[0-9]+]]:_(s16) = G_SELECT [[FCMP]](s1), [[C]], [[TRUNC]]
    ; GFX7-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[SELECT]](s16)
    ; GFX7-NEXT: $vgpr0 = COPY [[ANYEXT]](s32)
    ; GFX8-LABEL: name: test_fcmp_s16
    ; GFX8: liveins: $vgpr0
    ; GFX8-NEXT: {{  $}}
    ; GFX8-NEXT: [[C:%[0-9]+]]:_(s16) = G_CONSTANT i16 0
    ; GFX8-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
    ; GFX8-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY]](s32)
    ; GFX8-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s16), [[TRUNC]]
    ; GFX8-NEXT: [[SELECT:%[0-9]+]]:_(s16) = G_SELECT [[FCMP]](s1), [[C]], [[TRUNC]]
    ; GFX8-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[SELECT]](s16)
    ; GFX8-NEXT: $vgpr0 = COPY [[ANYEXT]](s32)
    ; GFX9-LABEL: name: test_fcmp_s16
    ; GFX9: liveins: $vgpr0
    ; GFX9-NEXT: {{  $}}
    ; GFX9-NEXT: [[C:%[0-9]+]]:_(s16) = G_CONSTANT i16 0
    ; GFX9-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
    ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY]](s32)
    ; GFX9-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s16), [[TRUNC]]
    ; GFX9-NEXT: [[SELECT:%[0-9]+]]:_(s16) = G_SELECT [[FCMP]](s1), [[C]], [[TRUNC]]
    ; GFX9-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[SELECT]](s16)
    ; GFX9-NEXT: $vgpr0 = COPY [[ANYEXT]](s32)
    %0:_(s16) = G_CONSTANT i16 0
    %1:_(s32) = COPY $vgpr0
    %2:_(s16) = G_TRUNC %1
    %3:_(s1) = G_FCMP floatpred(oeq), %0, %2
    %4:_(s16) = G_SELECT %3, %0, %2
    %5:_(s32) = G_ANYEXT %4
    $vgpr0 = COPY %5
...

---
name: test_fcmp_v2s32
body: |
  bb.0:
    liveins: $vgpr0_vgpr1
    ; GFX7-LABEL: name: test_fcmp_v2s32
    ; GFX7: liveins: $vgpr0_vgpr1
    ; GFX7-NEXT: {{  $}}
    ; GFX7-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
    ; GFX7-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
    ; GFX7-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
    ; GFX7-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s32), [[UV]]
    ; GFX7-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s32), [[UV1]]
    ; GFX7-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP]](s1)
    ; GFX7-NEXT: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP1]](s1)
    ; GFX7-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[ANYEXT]](s32), [[ANYEXT1]](s32)
    ; GFX7-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<2 x s32>)
    ; GFX8-LABEL: name: test_fcmp_v2s32
    ; GFX8: liveins: $vgpr0_vgpr1
    ; GFX8-NEXT: {{  $}}
    ; GFX8-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
    ; GFX8-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
    ; GFX8-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
    ; GFX8-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s32), [[UV]]
    ; GFX8-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s32), [[UV1]]
    ; GFX8-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP]](s1)
    ; GFX8-NEXT: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP1]](s1)
    ; GFX8-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[ANYEXT]](s32), [[ANYEXT1]](s32)
    ; GFX8-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<2 x s32>)
    ; GFX9-LABEL: name: test_fcmp_v2s32
    ; GFX9: liveins: $vgpr0_vgpr1
    ; GFX9-NEXT: {{  $}}
    ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
    ; GFX9-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
    ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
    ; GFX9-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s32), [[UV]]
    ; GFX9-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[C]](s32), [[UV1]]
    ; GFX9-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP]](s1)
    ; GFX9-NEXT: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP1]](s1)
    ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[ANYEXT]](s32), [[ANYEXT1]](s32)
    ; GFX9-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<2 x s32>)
    %0:_(s32) = G_CONSTANT i32 0
    %1:_(<2 x s32>) = G_BUILD_VECTOR %0, %0
    %2:_(<2 x s32>) = COPY $vgpr0_vgpr1
    %3:_(<2 x s1>) = G_FCMP floatpred(oeq), %1, %2
    %4:_(<2 x s32>) = G_ANYEXT %3
    S_NOP 0, implicit %4
...

---
name: test_fcmp_v2s32_flags
body: |
  bb.0:
    liveins: $vgpr0_vgpr1
    ; GFX7-LABEL: name: test_fcmp_v2s32_flags
    ; GFX7: liveins: $vgpr0_vgpr1
    ; GFX7-NEXT: {{  $}}
    ; GFX7-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
    ; GFX7-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
    ; GFX7-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
    ; GFX7-NEXT: [[FCMP:%[0-9]+]]:_(s1) = nnan G_FCMP floatpred(oeq), [[C]](s32), [[UV]]
    ; GFX7-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = nnan G_FCMP floatpred(oeq), [[C]](s32), [[UV1]]
    ; GFX7-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP]](s1)
    ; GFX7-NEXT: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP1]](s1)
    ; GFX7-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[ANYEXT]](s32), [[ANYEXT1]](s32)
    ; GFX7-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<2 x s32>)
    ; GFX8-LABEL: name: test_fcmp_v2s32_flags
    ; GFX8: liveins: $vgpr0_vgpr1
    ; GFX8-NEXT: {{  $}}
    ; GFX8-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
    ; GFX8-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
    ; GFX8-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
    ; GFX8-NEXT: [[FCMP:%[0-9]+]]:_(s1) = nnan G_FCMP floatpred(oeq), [[C]](s32), [[UV]]
    ; GFX8-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = nnan G_FCMP floatpred(oeq), [[C]](s32), [[UV1]]
    ; GFX8-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP]](s1)
    ; GFX8-NEXT: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP1]](s1)
    ; GFX8-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[ANYEXT]](s32), [[ANYEXT1]](s32)
    ; GFX8-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<2 x s32>)
    ; GFX9-LABEL: name: test_fcmp_v2s32_flags
    ; GFX9: liveins: $vgpr0_vgpr1
    ; GFX9-NEXT: {{  $}}
    ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
    ; GFX9-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
    ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
    ; GFX9-NEXT: [[FCMP:%[0-9]+]]:_(s1) = nnan G_FCMP floatpred(oeq), [[C]](s32), [[UV]]
    ; GFX9-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = nnan G_FCMP floatpred(oeq), [[C]](s32), [[UV1]]
    ; GFX9-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP]](s1)
    ; GFX9-NEXT: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP1]](s1)
    ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[ANYEXT]](s32), [[ANYEXT1]](s32)
    ; GFX9-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<2 x s32>)
    %0:_(s32) = G_CONSTANT i32 0
    %1:_(<2 x s32>) = G_BUILD_VECTOR %0, %0
    %2:_(<2 x s32>) = COPY $vgpr0_vgpr1
    %3:_(<2 x s1>) = nnan G_FCMP floatpred(oeq), %1, %2
    %4:_(<2 x s32>) = G_ANYEXT %3
    S_NOP 0, implicit %4
...

---
name: test_fcmp_v3s32
body: |
  bb.0:
    liveins: $vgpr0_vgpr1_vgpr2

    ; GFX7-LABEL: name: test_fcmp_v3s32
    ; GFX7: liveins: $vgpr0_vgpr1_vgpr2
    ; GFX7-NEXT: {{  $}}
    ; GFX7-NEXT: [[DEF:%[0-9]+]]:_(<3 x s32>) = G_IMPLICIT_DEF
    ; GFX7-NEXT: [[COPY:%[0-9]+]]:_(<3 x s32>) = COPY $vgpr0_vgpr1_vgpr2
    ; GFX7-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[DEF]](<3 x s32>)
    ; GFX7-NEXT: [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<3 x s32>)
    ; GFX7-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV]](s32), [[UV3]]
    ; GFX7-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV1]](s32), [[UV4]]
    ; GFX7-NEXT: [[FCMP2:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV2]](s32), [[UV5]]
    ; GFX7-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP]](s1)
    ; GFX7-NEXT: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP1]](s1)
    ; GFX7-NEXT: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP2]](s1)
    ; GFX7-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[ANYEXT]](s32), [[ANYEXT1]](s32), [[ANYEXT2]](s32)
    ; GFX7-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<3 x s32>)
    ; GFX8-LABEL: name: test_fcmp_v3s32
    ; GFX8: liveins: $vgpr0_vgpr1_vgpr2
    ; GFX8-NEXT: {{  $}}
    ; GFX8-NEXT: [[DEF:%[0-9]+]]:_(<3 x s32>) = G_IMPLICIT_DEF
    ; GFX8-NEXT: [[COPY:%[0-9]+]]:_(<3 x s32>) = COPY $vgpr0_vgpr1_vgpr2
    ; GFX8-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[DEF]](<3 x s32>)
    ; GFX8-NEXT: [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<3 x s32>)
    ; GFX8-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV]](s32), [[UV3]]
    ; GFX8-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV1]](s32), [[UV4]]
    ; GFX8-NEXT: [[FCMP2:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV2]](s32), [[UV5]]
    ; GFX8-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP]](s1)
    ; GFX8-NEXT: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP1]](s1)
    ; GFX8-NEXT: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP2]](s1)
    ; GFX8-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[ANYEXT]](s32), [[ANYEXT1]](s32), [[ANYEXT2]](s32)
    ; GFX8-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<3 x s32>)
    ; GFX9-LABEL: name: test_fcmp_v3s32
    ; GFX9: liveins: $vgpr0_vgpr1_vgpr2
    ; GFX9-NEXT: {{  $}}
    ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(<3 x s32>) = G_IMPLICIT_DEF
    ; GFX9-NEXT: [[COPY:%[0-9]+]]:_(<3 x s32>) = COPY $vgpr0_vgpr1_vgpr2
    ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[DEF]](<3 x s32>)
    ; GFX9-NEXT: [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<3 x s32>)
    ; GFX9-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV]](s32), [[UV3]]
    ; GFX9-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV1]](s32), [[UV4]]
    ; GFX9-NEXT: [[FCMP2:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV2]](s32), [[UV5]]
    ; GFX9-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP]](s1)
    ; GFX9-NEXT: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP1]](s1)
    ; GFX9-NEXT: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP2]](s1)
    ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[ANYEXT]](s32), [[ANYEXT1]](s32), [[ANYEXT2]](s32)
    ; GFX9-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<3 x s32>)
    %0:_(<3 x s32>) = G_IMPLICIT_DEF
    %1:_(<3 x s32>) = COPY $vgpr0_vgpr1_vgpr2
    %2:_(<3 x s1>) = G_FCMP floatpred(oeq), %0, %1
    %3:_(<3 x s32>) = G_ANYEXT %2
    S_NOP 0, implicit %3

...

---
name: test_fcmp_v4s32
body: |
  bb.0:
    liveins: $vgpr0_vgpr1_vgpr2_vgpr3

    ; GFX7-LABEL: name: test_fcmp_v4s32
    ; GFX7: liveins: $vgpr0_vgpr1_vgpr2_vgpr3
    ; GFX7-NEXT: {{  $}}
    ; GFX7-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
    ; GFX7-NEXT: [[LOAD:%[0-9]+]]:_(<4 x s32>) = G_LOAD [[DEF]](p1) :: (volatile load (<4 x s32>))
    ; GFX7-NEXT: [[COPY:%[0-9]+]]:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
    ; GFX7-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[LOAD]](<4 x s32>)
    ; GFX7-NEXT: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32), [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<4 x s32>)
    ; GFX7-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV]](s32), [[UV4]]
    ; GFX7-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV1]](s32), [[UV5]]
    ; GFX7-NEXT: [[FCMP2:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV2]](s32), [[UV6]]
    ; GFX7-NEXT: [[FCMP3:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV3]](s32), [[UV7]]
    ; GFX7-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP]](s1)
    ; GFX7-NEXT: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP1]](s1)
    ; GFX7-NEXT: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP2]](s1)
    ; GFX7-NEXT: [[ANYEXT3:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP3]](s1)
    ; GFX7-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[ANYEXT]](s32), [[ANYEXT1]](s32), [[ANYEXT2]](s32), [[ANYEXT3]](s32)
    ; GFX7-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<4 x s32>)
    ; GFX8-LABEL: name: test_fcmp_v4s32
    ; GFX8: liveins: $vgpr0_vgpr1_vgpr2_vgpr3
    ; GFX8-NEXT: {{  $}}
    ; GFX8-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
    ; GFX8-NEXT: [[LOAD:%[0-9]+]]:_(<4 x s32>) = G_LOAD [[DEF]](p1) :: (volatile load (<4 x s32>))
    ; GFX8-NEXT: [[COPY:%[0-9]+]]:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
    ; GFX8-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[LOAD]](<4 x s32>)
    ; GFX8-NEXT: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32), [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<4 x s32>)
    ; GFX8-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV]](s32), [[UV4]]
    ; GFX8-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV1]](s32), [[UV5]]
    ; GFX8-NEXT: [[FCMP2:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV2]](s32), [[UV6]]
    ; GFX8-NEXT: [[FCMP3:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV3]](s32), [[UV7]]
    ; GFX8-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP]](s1)
    ; GFX8-NEXT: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP1]](s1)
    ; GFX8-NEXT: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP2]](s1)
    ; GFX8-NEXT: [[ANYEXT3:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP3]](s1)
    ; GFX8-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[ANYEXT]](s32), [[ANYEXT1]](s32), [[ANYEXT2]](s32), [[ANYEXT3]](s32)
    ; GFX8-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<4 x s32>)
    ; GFX9-LABEL: name: test_fcmp_v4s32
    ; GFX9: liveins: $vgpr0_vgpr1_vgpr2_vgpr3
    ; GFX9-NEXT: {{  $}}
    ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
    ; GFX9-NEXT: [[LOAD:%[0-9]+]]:_(<4 x s32>) = G_LOAD [[DEF]](p1) :: (volatile load (<4 x s32>))
    ; GFX9-NEXT: [[COPY:%[0-9]+]]:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
    ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[LOAD]](<4 x s32>)
    ; GFX9-NEXT: [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32), [[UV6:%[0-9]+]]:_(s32), [[UV7:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<4 x s32>)
    ; GFX9-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV]](s32), [[UV4]]
    ; GFX9-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV1]](s32), [[UV5]]
    ; GFX9-NEXT: [[FCMP2:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV2]](s32), [[UV6]]
    ; GFX9-NEXT: [[FCMP3:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[UV3]](s32), [[UV7]]
    ; GFX9-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP]](s1)
    ; GFX9-NEXT: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP1]](s1)
    ; GFX9-NEXT: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP2]](s1)
    ; GFX9-NEXT: [[ANYEXT3:%[0-9]+]]:_(s32) = G_ANYEXT [[FCMP3]](s1)
    ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[ANYEXT]](s32), [[ANYEXT1]](s32), [[ANYEXT2]](s32), [[ANYEXT3]](s32)
    ; GFX9-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<4 x s32>)
    %0:_(p1) = G_IMPLICIT_DEF
    %1:_(<4 x s32>) = G_LOAD %0 :: (volatile load (<4 x s32>))
    %2:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
    %3:_(<4 x s1>) = G_FCMP floatpred(oeq) , %1, %2
    %4:_(<4 x s32>) = G_ANYEXT %3
    S_NOP 0, implicit %4

...

---
name: test_icmp_v2s16
body: |
  bb.0:
    liveins: $vgpr0, $vgpr1, $vgpr2_vgpr3, $vgpr4_vgpr5
    ; GFX7-LABEL: name: test_icmp_v2s16
    ; GFX7: liveins: $vgpr0, $vgpr1, $vgpr2_vgpr3, $vgpr4_vgpr5
    ; GFX7-NEXT: {{  $}}
    ; GFX7-NEXT: [[COPY:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
    ; GFX7-NEXT: [[COPY1:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1
    ; GFX7-NEXT: [[COPY2:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr2_vgpr3
    ; GFX7-NEXT: [[COPY3:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr4_vgpr5
    ; GFX7-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY]](<2 x s16>)
    ; GFX7-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST]](s32)
    ; GFX7-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
    ; GFX7-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
    ; GFX7-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
    ; GFX7-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY1]](<2 x s16>)
    ; GFX7-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
    ; GFX7-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C]](s32)
    ; GFX7-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
    ; GFX7-NEXT: [[FPEXT:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC]](s16)
    ; GFX7-NEXT: [[FPEXT1:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC2]](s16)
    ; GFX7-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[FPEXT]](s32), [[FPEXT1]]
    ; GFX7-NEXT: [[FPEXT2:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC1]](s16)
    ; GFX7-NEXT: [[FPEXT3:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC3]](s16)
    ; GFX7-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[FPEXT2]](s32), [[FPEXT3]]
    ; GFX7-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY2]](<2 x s32>)
    ; GFX7-NEXT: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY3]](<2 x s32>)
    ; GFX7-NEXT: [[SELECT:%[0-9]+]]:_(s32) = G_SELECT [[FCMP]](s1), [[UV]], [[UV2]]
    ; GFX7-NEXT: [[SELECT1:%[0-9]+]]:_(s32) = G_SELECT [[FCMP1]](s1), [[UV1]], [[UV3]]
    ; GFX7-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32)
    ; GFX7-NEXT: $vgpr0_vgpr1 = COPY [[BUILD_VECTOR]](<2 x s32>)
    ; GFX8-LABEL: name: test_icmp_v2s16
    ; GFX8: liveins: $vgpr0, $vgpr1, $vgpr2_vgpr3, $vgpr4_vgpr5
    ; GFX8-NEXT: {{  $}}
    ; GFX8-NEXT: [[COPY:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
    ; GFX8-NEXT: [[COPY1:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1
    ; GFX8-NEXT: [[COPY2:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr2_vgpr3
    ; GFX8-NEXT: [[COPY3:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr4_vgpr5
    ; GFX8-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY]](<2 x s16>)
    ; GFX8-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST]](s32)
    ; GFX8-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
    ; GFX8-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
    ; GFX8-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
    ; GFX8-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY1]](<2 x s16>)
    ; GFX8-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
    ; GFX8-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C]](s32)
    ; GFX8-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
    ; GFX8-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[TRUNC]](s16), [[TRUNC2]]
    ; GFX8-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[TRUNC1]](s16), [[TRUNC3]]
    ; GFX8-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY2]](<2 x s32>)
    ; GFX8-NEXT: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY3]](<2 x s32>)
    ; GFX8-NEXT: [[SELECT:%[0-9]+]]:_(s32) = G_SELECT [[FCMP]](s1), [[UV]], [[UV2]]
    ; GFX8-NEXT: [[SELECT1:%[0-9]+]]:_(s32) = G_SELECT [[FCMP1]](s1), [[UV1]], [[UV3]]
    ; GFX8-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32)
    ; GFX8-NEXT: $vgpr0_vgpr1 = COPY [[BUILD_VECTOR]](<2 x s32>)
    ; GFX9-LABEL: name: test_icmp_v2s16
    ; GFX9: liveins: $vgpr0, $vgpr1, $vgpr2_vgpr3, $vgpr4_vgpr5
    ; GFX9-NEXT: {{  $}}
    ; GFX9-NEXT: [[COPY:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
    ; GFX9-NEXT: [[COPY1:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1
    ; GFX9-NEXT: [[COPY2:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr2_vgpr3
    ; GFX9-NEXT: [[COPY3:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr4_vgpr5
    ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY]](<2 x s16>)
    ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST]](s32)
    ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
    ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
    ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
    ; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY1]](<2 x s16>)
    ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
    ; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C]](s32)
    ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
    ; GFX9-NEXT: [[FCMP:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[TRUNC]](s16), [[TRUNC2]]
    ; GFX9-NEXT: [[FCMP1:%[0-9]+]]:_(s1) = G_FCMP floatpred(oeq), [[TRUNC1]](s16), [[TRUNC3]]
    ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY2]](<2 x s32>)
    ; GFX9-NEXT: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY3]](<2 x s32>)
    ; GFX9-NEXT: [[SELECT:%[0-9]+]]:_(s32) = G_SELECT [[FCMP]](s1), [[UV]], [[UV2]]
    ; GFX9-NEXT: [[SELECT1:%[0-9]+]]:_(s32) = G_SELECT [[FCMP1]](s1), [[UV1]], [[UV3]]
    ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32)
    ; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[BUILD_VECTOR]](<2 x s32>)
    %0:_(<2 x s16>) = COPY $vgpr0
    %1:_(<2 x s16>) = COPY $vgpr1
    %2:_(<2 x s32>) = COPY $vgpr2_vgpr3
    %3:_(<2 x s32>) = COPY $vgpr4_vgpr5
    %4:_(<2 x s1>) = G_FCMP floatpred(oeq), %0, %1
    %5:_(<2 x s32>) = G_SELECT %4, %2, %3
    $vgpr0_vgpr1 = COPY %5
...