llvm/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-strided-load-store-asm.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+v,+zvl256b | FileCheck %s --check-prefixes=CHECK,V
; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+f,+zve32f,+zvl256b | FileCheck %s --check-prefixes=CHECK,ZVE32F
; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+v,+optimized-zero-stride-load,+zvl256b | FileCheck %s --check-prefixes=CHECK,OPTIMIZED
; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+f,+zve32f,+optimized-zero-stride-load,+zvl256b | FileCheck %s --check-prefixes=CHECK,OPTIMIZED

%struct.foo = type { i32, i32, i32, i32 }

; void gather(signed char * __restrict  A, signed char * __restrict B) {
;   for (int i = 0; i != 1024; ++i)
;       A[i] += B[i * 5];
; }
define void @gather(ptr noalias nocapture %A, ptr noalias nocapture readonly %B) {
; CHECK-LABEL: gather:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    addi a2, a0, 1024
; CHECK-NEXT:    li a4, 32
; CHECK-NEXT:    li a3, 5
; CHECK-NEXT:    vsetvli zero, a4, e8, m1, ta, ma
; CHECK-NEXT:  .LBB0_1: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    vlse8.v v8, (a1), a3
; CHECK-NEXT:    vle8.v v9, (a0)
; CHECK-NEXT:    vadd.vv v8, v9, v8
; CHECK-NEXT:    vse8.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 32
; CHECK-NEXT:    addi a1, a1, 160
; CHECK-NEXT:    bne a0, a2, .LBB0_1
; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
; CHECK-NEXT:    ret
entry:
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.ind = phi <32 x i64> [ <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15, i64 16, i64 17, i64 18, i64 19, i64 20, i64 21, i64 22, i64 23, i64 24, i64 25, i64 26, i64 27, i64 28, i64 29, i64 30, i64 31>, %entry ], [ %vec.ind.next, %vector.body ]
  %i = mul nuw nsw <32 x i64> %vec.ind, splat (i64 5)
  %i1 = getelementptr inbounds i8, ptr %B, <32 x i64> %i
  %wide.masked.gather = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> %i1, i32 1, <32 x i1> splat (i1 true), <32 x i8> undef)
  %i2 = getelementptr inbounds i8, ptr %A, i64 %index
  %wide.load = load <32 x i8>, ptr %i2, align 1
  %i4 = add <32 x i8> %wide.load, %wide.masked.gather
  store <32 x i8> %i4, ptr %i2, align 1
  %index.next = add nuw i64 %index, 32
  %vec.ind.next = add <32 x i64> %vec.ind, splat (i64 32)
  %i6 = icmp eq i64 %index.next, 1024
  br i1 %i6, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}

define void @gather_masked(ptr noalias nocapture %A, ptr noalias nocapture readonly %B, <32 x i8> %maskedoff) {
; CHECK-LABEL: gather_masked:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    addi a2, a0, 1024
; CHECK-NEXT:    lui a3, 983765
; CHECK-NEXT:    addi a3, a3, 873
; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
; CHECK-NEXT:    vmv.s.x v0, a3
; CHECK-NEXT:    li a3, 32
; CHECK-NEXT:    li a4, 5
; CHECK-NEXT:  .LBB1_1: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    vmv1r.v v9, v8
; CHECK-NEXT:    vsetvli zero, a3, e8, m1, ta, mu
; CHECK-NEXT:    vlse8.v v9, (a1), a4, v0.t
; CHECK-NEXT:    vle8.v v10, (a0)
; CHECK-NEXT:    vadd.vv v9, v10, v9
; CHECK-NEXT:    vse8.v v9, (a0)
; CHECK-NEXT:    addi a0, a0, 32
; CHECK-NEXT:    addi a1, a1, 160
; CHECK-NEXT:    bne a0, a2, .LBB1_1
; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
; CHECK-NEXT:    ret
entry:
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.ind = phi <32 x i64> [ <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15, i64 16, i64 17, i64 18, i64 19, i64 20, i64 21, i64 22, i64 23, i64 24, i64 25, i64 26, i64 27, i64 28, i64 29, i64 30, i64 31>, %entry ], [ %vec.ind.next, %vector.body ]
  %i = mul nuw nsw <32 x i64> %vec.ind, splat (i64 5)
  %i1 = getelementptr inbounds i8, ptr %B, <32 x i64> %i
  %wide.masked.gather = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> %i1, i32 1, <32 x i1> <i1 true, i1 false, i1 false, i1 true, i1 false, i1 true, i1 true, i1 false, i1 true, i1 true, i1 false, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 true, i1 false, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true>, <32 x i8> %maskedoff)
  %i2 = getelementptr inbounds i8, ptr %A, i64 %index
  %wide.load = load <32 x i8>, ptr %i2, align 1
  %i4 = add <32 x i8> %wide.load, %wide.masked.gather
  store <32 x i8> %i4, ptr %i2, align 1
  %index.next = add nuw i64 %index, 32
  %vec.ind.next = add <32 x i64> %vec.ind, splat (i64 32)
  %i6 = icmp eq i64 %index.next, 1024
  br i1 %i6, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}

define void @gather_negative_stride(ptr noalias nocapture %A, ptr noalias nocapture readonly %B) {
; CHECK-LABEL: gather_negative_stride:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    addi a1, a1, 155
; CHECK-NEXT:    addi a2, a0, 1024
; CHECK-NEXT:    li a4, 32
; CHECK-NEXT:    li a3, -5
; CHECK-NEXT:    vsetvli zero, a4, e8, m1, ta, ma
; CHECK-NEXT:  .LBB2_1: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    vlse8.v v8, (a1), a3
; CHECK-NEXT:    vle8.v v9, (a0)
; CHECK-NEXT:    vadd.vv v8, v9, v8
; CHECK-NEXT:    vse8.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 32
; CHECK-NEXT:    addi a1, a1, 160
; CHECK-NEXT:    bne a0, a2, .LBB2_1
; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
; CHECK-NEXT:    ret
entry:
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.ind = phi <32 x i64> [ <i64 31, i64 30, i64 29, i64 28, i64 27, i64 26, i64 25, i64 24, i64 23, i64 22, i64 21, i64 20, i64 19, i64 18, i64 17, i64 16, i64 15, i64 14, i64 13, i64 12, i64 11, i64 10, i64 9, i64 8, i64 7, i64 6, i64 5, i64 4, i64 3, i64 2, i64 1, i64 0>, %entry ], [ %vec.ind.next, %vector.body ]
  %i = mul nuw nsw <32 x i64> %vec.ind, splat (i64 5)
  %i1 = getelementptr inbounds i8, ptr %B, <32 x i64> %i
  %wide.masked.gather = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> %i1, i32 1, <32 x i1> splat (i1 true), <32 x i8> undef)
  %i2 = getelementptr inbounds i8, ptr %A, i64 %index
  %wide.load = load <32 x i8>, ptr %i2, align 1
  %i4 = add <32 x i8> %wide.load, %wide.masked.gather
  store <32 x i8> %i4, ptr %i2, align 1
  %index.next = add nuw i64 %index, 32
  %vec.ind.next = add <32 x i64> %vec.ind, splat (i64 32)
  %i6 = icmp eq i64 %index.next, 1024
  br i1 %i6, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}

define void @gather_zero_stride(ptr noalias nocapture %A, ptr noalias nocapture readonly %B) {
; CHECK-LABEL: gather_zero_stride:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    addi a2, a0, 1024
; CHECK-NEXT:    li a3, 32
; CHECK-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
; CHECK-NEXT:  .LBB3_1: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    lbu a3, 0(a1)
; CHECK-NEXT:    vle8.v v8, (a0)
; CHECK-NEXT:    vadd.vx v8, v8, a3
; CHECK-NEXT:    vse8.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 32
; CHECK-NEXT:    addi a1, a1, 160
; CHECK-NEXT:    bne a0, a2, .LBB3_1
; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
; CHECK-NEXT:    ret
entry:
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.ind = phi <32 x i64> [ zeroinitializer, %entry ], [ %vec.ind.next, %vector.body ]
  %i = mul nuw nsw <32 x i64> %vec.ind, splat (i64 5)
  %i1 = getelementptr inbounds i8, ptr %B, <32 x i64> %i
  %wide.masked.gather = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> %i1, i32 1, <32 x i1> splat (i1 true), <32 x i8> undef)
  %i2 = getelementptr inbounds i8, ptr %A, i64 %index
  %wide.load = load <32 x i8>, ptr %i2, align 1
  %i4 = add <32 x i8> %wide.load, %wide.masked.gather
  store <32 x i8> %i4, ptr %i2, align 1
  %index.next = add nuw i64 %index, 32
  %vec.ind.next = add <32 x i64> %vec.ind, splat (i64 32)
  %i6 = icmp eq i64 %index.next, 1024
  br i1 %i6, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}

define void @gather_zero_stride_i32(ptr noalias nocapture %A, ptr noalias nocapture readonly %B) {
; CHECK-LABEL: gather_zero_stride_i32:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    addi a2, a0, 1024
; CHECK-NEXT:    vsetivli zero, 8, e32, m1, ta, ma
; CHECK-NEXT:  .LBB4_1: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    lw a3, 0(a1)
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vadd.vx v8, v8, a3
; CHECK-NEXT:    vse32.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 8
; CHECK-NEXT:    addi a1, a1, 160
; CHECK-NEXT:    bne a0, a2, .LBB4_1
; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
; CHECK-NEXT:    ret
entry:
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.ind = phi <8 x i64> [ zeroinitializer, %entry ], [ %vec.ind.next, %vector.body ]
  %i = mul nuw nsw <8 x i64> %vec.ind, splat (i64 5)
  %i1 = getelementptr inbounds i8, ptr %B, <8 x i64> %i
  %wide.masked.gather = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i1, i32 4, <8 x i1> splat (i1 true), <8 x i32> undef)
  %i2 = getelementptr inbounds i8, ptr %A, i64 %index
  %wide.load = load <8 x i32>, ptr %i2, align 4
  %i4 = add <8 x i32> %wide.load, %wide.masked.gather
  store <8 x i32> %i4, ptr %i2, align 4
  %index.next = add nuw i64 %index, 8
  %vec.ind.next = add <8 x i64> %vec.ind, splat (i64 32)
  %i6 = icmp eq i64 %index.next, 1024
  br i1 %i6, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}

define void @gather_zero_stride_unfold(ptr noalias nocapture %A, ptr noalias nocapture readonly %B) {
; V-LABEL: gather_zero_stride_unfold:
; V:       # %bb.0: # %entry
; V-NEXT:    addi a2, a0, 1024
; V-NEXT:    li a3, 32
; V-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
; V-NEXT:  .LBB5_1: # %vector.body
; V-NEXT:    # =>This Inner Loop Header: Depth=1
; V-NEXT:    lbu a3, 0(a1)
; V-NEXT:    vle8.v v8, (a0)
; V-NEXT:    vmv.v.x v9, a3
; V-NEXT:    vdivu.vv v8, v9, v8
; V-NEXT:    vse8.v v8, (a0)
; V-NEXT:    addi a0, a0, 32
; V-NEXT:    addi a1, a1, 160
; V-NEXT:    bne a0, a2, .LBB5_1
; V-NEXT:  # %bb.2: # %for.cond.cleanup
; V-NEXT:    ret
;
; ZVE32F-LABEL: gather_zero_stride_unfold:
; ZVE32F:       # %bb.0: # %entry
; ZVE32F-NEXT:    addi a2, a0, 1024
; ZVE32F-NEXT:    li a3, 32
; ZVE32F-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
; ZVE32F-NEXT:  .LBB5_1: # %vector.body
; ZVE32F-NEXT:    # =>This Inner Loop Header: Depth=1
; ZVE32F-NEXT:    lbu a3, 0(a1)
; ZVE32F-NEXT:    vle8.v v8, (a0)
; ZVE32F-NEXT:    vmv.v.x v9, a3
; ZVE32F-NEXT:    vdivu.vv v8, v9, v8
; ZVE32F-NEXT:    vse8.v v8, (a0)
; ZVE32F-NEXT:    addi a0, a0, 32
; ZVE32F-NEXT:    addi a1, a1, 160
; ZVE32F-NEXT:    bne a0, a2, .LBB5_1
; ZVE32F-NEXT:  # %bb.2: # %for.cond.cleanup
; ZVE32F-NEXT:    ret
;
; OPTIMIZED-LABEL: gather_zero_stride_unfold:
; OPTIMIZED:       # %bb.0: # %entry
; OPTIMIZED-NEXT:    addi a2, a0, 1024
; OPTIMIZED-NEXT:    li a3, 32
; OPTIMIZED-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
; OPTIMIZED-NEXT:  .LBB5_1: # %vector.body
; OPTIMIZED-NEXT:    # =>This Inner Loop Header: Depth=1
; OPTIMIZED-NEXT:    vlse8.v v8, (a1), zero
; OPTIMIZED-NEXT:    vle8.v v9, (a0)
; OPTIMIZED-NEXT:    vdivu.vv v8, v8, v9
; OPTIMIZED-NEXT:    vse8.v v8, (a0)
; OPTIMIZED-NEXT:    addi a0, a0, 32
; OPTIMIZED-NEXT:    addi a1, a1, 160
; OPTIMIZED-NEXT:    bne a0, a2, .LBB5_1
; OPTIMIZED-NEXT:  # %bb.2: # %for.cond.cleanup
; OPTIMIZED-NEXT:    ret
entry:
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.ind = phi <32 x i64> [ zeroinitializer, %entry ], [ %vec.ind.next, %vector.body ]
  %i = mul nuw nsw <32 x i64> %vec.ind, splat (i64 5)
  %i1 = getelementptr inbounds i8, ptr %B, <32 x i64> %i
  %wide.masked.gather = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> %i1, i32 1, <32 x i1> splat (i1 true), <32 x i8> undef)
  %i2 = getelementptr inbounds i8, ptr %A, i64 %index
  %wide.load = load <32 x i8>, ptr %i2, align 1
  %i4 = udiv <32 x i8> %wide.masked.gather, %wide.load
  store <32 x i8> %i4, ptr %i2, align 1
  %index.next = add nuw i64 %index, 32
  %vec.ind.next = add <32 x i64> %vec.ind, splat (i64 32)
  %i6 = icmp eq i64 %index.next, 1024
  br i1 %i6, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}

;void scatter(signed char * __restrict  A, signed char * __restrict B) {
;  for (int i = 0; i < 1024; ++i)
;      A[i * 5] += B[i];
;}
define void @scatter(ptr noalias nocapture %A, ptr noalias nocapture readonly %B) {
; CHECK-LABEL: scatter:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    addi a2, a1, 1024
; CHECK-NEXT:    li a4, 32
; CHECK-NEXT:    li a3, 5
; CHECK-NEXT:    vsetvli zero, a4, e8, m1, ta, ma
; CHECK-NEXT:  .LBB6_1: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    vle8.v v8, (a1)
; CHECK-NEXT:    vlse8.v v9, (a0), a3
; CHECK-NEXT:    vadd.vv v8, v9, v8
; CHECK-NEXT:    vsse8.v v8, (a0), a3
; CHECK-NEXT:    addi a1, a1, 32
; CHECK-NEXT:    addi a0, a0, 160
; CHECK-NEXT:    bne a1, a2, .LBB6_1
; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
; CHECK-NEXT:    ret
entry:
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.ind = phi <32 x i64> [ <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15, i64 16, i64 17, i64 18, i64 19, i64 20, i64 21, i64 22, i64 23, i64 24, i64 25, i64 26, i64 27, i64 28, i64 29, i64 30, i64 31>, %entry ], [ %vec.ind.next, %vector.body ]
  %i = getelementptr inbounds i8, ptr %B, i64 %index
  %wide.load = load <32 x i8>, ptr %i, align 1
  %i2 = mul nuw nsw <32 x i64> %vec.ind, splat (i64 5)
  %i3 = getelementptr inbounds i8, ptr %A, <32 x i64> %i2
  %wide.masked.gather = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> %i3, i32 1, <32 x i1> splat (i1 true), <32 x i8> undef)
  %i4 = add <32 x i8> %wide.masked.gather, %wide.load
  call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> %i4, <32 x ptr> %i3, i32 1, <32 x i1> splat (i1 true))
  %index.next = add nuw i64 %index, 32
  %vec.ind.next = add <32 x i64> %vec.ind, splat (i64 32)
  %i5 = icmp eq i64 %index.next, 1024
  br i1 %i5, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}

define void @scatter_masked(ptr noalias nocapture %A, ptr noalias nocapture readonly %B, <32 x i8> %maskedoff) {
; CHECK-LABEL: scatter_masked:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    addi a2, a1, 1024
; CHECK-NEXT:    li a3, 32
; CHECK-NEXT:    lui a4, 983765
; CHECK-NEXT:    addi a4, a4, 873
; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
; CHECK-NEXT:    vmv.s.x v0, a4
; CHECK-NEXT:    li a4, 5
; CHECK-NEXT:  .LBB7_1: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    vsetvli zero, a3, e8, m1, ta, mu
; CHECK-NEXT:    vle8.v v9, (a1)
; CHECK-NEXT:    vmv1r.v v10, v8
; CHECK-NEXT:    vlse8.v v10, (a0), a4, v0.t
; CHECK-NEXT:    vadd.vv v9, v10, v9
; CHECK-NEXT:    vsse8.v v9, (a0), a4, v0.t
; CHECK-NEXT:    addi a1, a1, 32
; CHECK-NEXT:    addi a0, a0, 160
; CHECK-NEXT:    bne a1, a2, .LBB7_1
; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
; CHECK-NEXT:    ret
entry:
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.ind = phi <32 x i64> [ <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15, i64 16, i64 17, i64 18, i64 19, i64 20, i64 21, i64 22, i64 23, i64 24, i64 25, i64 26, i64 27, i64 28, i64 29, i64 30, i64 31>, %entry ], [ %vec.ind.next, %vector.body ]
  %i = getelementptr inbounds i8, ptr %B, i64 %index
  %wide.load = load <32 x i8>, ptr %i, align 1
  %i2 = mul nuw nsw <32 x i64> %vec.ind, splat (i64 5)
  %i3 = getelementptr inbounds i8, ptr %A, <32 x i64> %i2
  %wide.masked.gather = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> %i3, i32 1, <32 x i1> <i1 true, i1 false, i1 false, i1 true, i1 false, i1 true, i1 true, i1 false, i1 true, i1 true, i1 false, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 true, i1 false, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true>, <32 x i8> %maskedoff)
  %i4 = add <32 x i8> %wide.masked.gather, %wide.load
  call void @llvm.masked.scatter.v32i8.v32p0(<32 x i8> %i4, <32 x ptr> %i3, i32 1, <32 x i1> <i1 true, i1 false, i1 false, i1 true, i1 false, i1 true, i1 true, i1 false, i1 true, i1 true, i1 false, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 true, i1 false, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true>)
  %index.next = add nuw i64 %index, 32
  %vec.ind.next = add <32 x i64> %vec.ind, splat (i64 32)
  %i5 = icmp eq i64 %index.next, 1024
  br i1 %i5, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}

; void gather_pow2(signed char * __restrict  A, signed char * __restrict B) {
;   for (int i = 0; i != 1024; ++i)
;       A[i] += B[i * 4];
; }
define void @gather_pow2(ptr noalias nocapture %A, ptr noalias nocapture readonly %B) {
; CHECK-LABEL: gather_pow2:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    lui a2, 1
; CHECK-NEXT:    add a2, a0, a2
; CHECK-NEXT:    li a3, 16
; CHECK-NEXT:    li a4, 32
; CHECK-NEXT:  .LBB8_1: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    vsetivli zero, 8, e32, m1, ta, ma
; CHECK-NEXT:    vlse32.v v8, (a1), a3
; CHECK-NEXT:    vsetvli zero, a4, e8, m1, ta, ma
; CHECK-NEXT:    vle8.v v9, (a0)
; CHECK-NEXT:    vsetivli zero, 8, e32, m1, ta, ma
; CHECK-NEXT:    vadd.vv v8, v9, v8
; CHECK-NEXT:    vsetvli zero, a4, e8, m1, ta, ma
; CHECK-NEXT:    vse8.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 32
; CHECK-NEXT:    addi a1, a1, 128
; CHECK-NEXT:    bne a0, a2, .LBB8_1
; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
; CHECK-NEXT:    ret
entry:
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.ind = phi <8 x i64> [ <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7>, %entry ], [ %vec.ind.next, %vector.body ]
  %i = shl nsw <8 x i64> %vec.ind, splat (i64 2)
  %i1 = getelementptr inbounds i32, ptr %B, <8 x i64> %i
  %wide.masked.gather = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i1, i32 4, <8 x i1> splat (i1 true), <8 x i32> undef)
  %i2 = getelementptr inbounds i32, ptr %A, i64 %index
  %wide.load = load <8 x i32>, ptr %i2, align 1
  %i4 = add <8 x i32> %wide.load, %wide.masked.gather
  store <8 x i32> %i4, ptr %i2, align 1
  %index.next = add nuw i64 %index, 8
  %vec.ind.next = add <8 x i64> %vec.ind, splat (i64 8)
  %i6 = icmp eq i64 %index.next, 1024
  br i1 %i6, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}

;void scatter_pow2(signed char * __restrict  A, signed char * __restrict B) {
;  for (int i = 0; i < 1024; ++i)
;      A[i * 4] += B[i];
;}
define void @scatter_pow2(ptr noalias nocapture %A, ptr noalias nocapture readonly %B) {
; CHECK-LABEL: scatter_pow2:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    lui a2, 1
; CHECK-NEXT:    add a2, a1, a2
; CHECK-NEXT:    li a3, 32
; CHECK-NEXT:    li a4, 16
; CHECK-NEXT:  .LBB9_1: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    vsetvli zero, a3, e8, m1, ta, ma
; CHECK-NEXT:    vle8.v v8, (a1)
; CHECK-NEXT:    vsetivli zero, 8, e32, m1, ta, ma
; CHECK-NEXT:    vlse32.v v9, (a0), a4
; CHECK-NEXT:    vadd.vv v8, v9, v8
; CHECK-NEXT:    vsse32.v v8, (a0), a4
; CHECK-NEXT:    addi a1, a1, 32
; CHECK-NEXT:    addi a0, a0, 128
; CHECK-NEXT:    bne a1, a2, .LBB9_1
; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
; CHECK-NEXT:    ret
entry:
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.ind = phi <8 x i64> [ <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7>, %entry ], [ %vec.ind.next, %vector.body ]
  %i = getelementptr inbounds i32, ptr %B, i64 %index
  %wide.load = load <8 x i32>, ptr %i, align 1
  %i2 = shl nuw nsw <8 x i64> %vec.ind, splat (i64 2)
  %i3 = getelementptr inbounds i32, ptr %A, <8 x i64> %i2
  %wide.masked.gather = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i3, i32 4, <8 x i1> splat (i1 true), <8 x i32> undef)
  %i4 = add <8 x i32> %wide.masked.gather, %wide.load
  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> %i4, <8 x ptr> %i3, i32 4, <8 x i1> splat (i1 true))
  %index.next = add nuw i64 %index, 8
  %vec.ind.next = add <8 x i64> %vec.ind, splat (i64 8)
  %i5 = icmp eq i64 %index.next, 1024
  br i1 %i5, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}

;struct foo {
;  int a, b, c, d;
;};
;
;void struct_gather(int * __restrict  A, struct foo * __restrict B) {
;  for (int i = 0; i < 1024; ++i)
;      A[i] += B[i].b;
;}
define void @struct_gather(ptr noalias nocapture %A, ptr noalias nocapture readonly %B) {
; CHECK-LABEL: struct_gather:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    addi a1, a1, 132
; CHECK-NEXT:    lui a2, 1
; CHECK-NEXT:    add a2, a0, a2
; CHECK-NEXT:    li a3, 16
; CHECK-NEXT:    vsetivli zero, 8, e32, m1, ta, ma
; CHECK-NEXT:  .LBB10_1: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    addi a4, a0, 32
; CHECK-NEXT:    addi a5, a1, -128
; CHECK-NEXT:    vlse32.v v8, (a5), a3
; CHECK-NEXT:    vlse32.v v9, (a1), a3
; CHECK-NEXT:    vle32.v v10, (a0)
; CHECK-NEXT:    vle32.v v11, (a4)
; CHECK-NEXT:    vadd.vv v8, v10, v8
; CHECK-NEXT:    vadd.vv v9, v11, v9
; CHECK-NEXT:    vse32.v v8, (a0)
; CHECK-NEXT:    vse32.v v9, (a4)
; CHECK-NEXT:    addi a0, a0, 64
; CHECK-NEXT:    addi a1, a1, 256
; CHECK-NEXT:    bne a0, a2, .LBB10_1
; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
; CHECK-NEXT:    ret
entry:
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.ind = phi <8 x i64> [ <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7>, %entry ], [ %vec.ind.next, %vector.body ]
  %step.add = add <8 x i64> %vec.ind, splat (i64 8)
  %i = getelementptr inbounds %struct.foo, ptr %B, <8 x i64> %vec.ind, i32 1
  %i1 = getelementptr inbounds %struct.foo, ptr %B, <8 x i64> %step.add, i32 1
  %wide.masked.gather = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i, i32 4, <8 x i1> splat (i1 true), <8 x i32> undef)
  %wide.masked.gather9 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i1, i32 4, <8 x i1> splat (i1 true), <8 x i32> undef)
  %i2 = getelementptr inbounds i32, ptr %A, i64 %index
  %wide.load = load <8 x i32>, ptr %i2, align 4
  %i4 = getelementptr inbounds i32, ptr %i2, i64 8
  %wide.load10 = load <8 x i32>, ptr %i4, align 4
  %i6 = add nsw <8 x i32> %wide.load, %wide.masked.gather
  %i7 = add nsw <8 x i32> %wide.load10, %wide.masked.gather9
  store <8 x i32> %i6, ptr %i2, align 4
  store <8 x i32> %i7, ptr %i4, align 4
  %index.next = add nuw i64 %index, 16
  %vec.ind.next = add <8 x i64> %vec.ind, splat (i64 16)
  %i10 = icmp eq i64 %index.next, 1024
  br i1 %i10, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}

;void gather_unroll(int * __restrict  A, int * __restrict B) {
;  for (int i = 0; i < 1024; i+= 4 ) {
;    A[i] += B[i * 4];
;    A[i+1] += B[(i+1) * 4];
;    A[i+2] += B[(i+2) * 4];
;    A[i+3] += B[(i+3) * 4];
;  }
;}
define void @gather_unroll(ptr noalias nocapture %A, ptr noalias nocapture readonly %B) {
; CHECK-LABEL: gather_unroll:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    li a2, 256
; CHECK-NEXT:    li a3, 64
; CHECK-NEXT:    li a4, 16
; CHECK-NEXT:    vsetivli zero, 8, e32, m1, ta, ma
; CHECK-NEXT:  .LBB11_1: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    vlse32.v v8, (a1), a3
; CHECK-NEXT:    vlse32.v v9, (a0), a4
; CHECK-NEXT:    vadd.vv v8, v9, v8
; CHECK-NEXT:    vsse32.v v8, (a0), a4
; CHECK-NEXT:    addi a5, a1, 16
; CHECK-NEXT:    vlse32.v v8, (a5), a3
; CHECK-NEXT:    addi a5, a0, 4
; CHECK-NEXT:    vlse32.v v9, (a5), a4
; CHECK-NEXT:    vadd.vv v8, v9, v8
; CHECK-NEXT:    vsse32.v v8, (a5), a4
; CHECK-NEXT:    addi a5, a1, 32
; CHECK-NEXT:    vlse32.v v8, (a5), a3
; CHECK-NEXT:    addi a5, a0, 8
; CHECK-NEXT:    vlse32.v v9, (a5), a4
; CHECK-NEXT:    vadd.vv v8, v9, v8
; CHECK-NEXT:    vsse32.v v8, (a5), a4
; CHECK-NEXT:    addi a5, a1, 48
; CHECK-NEXT:    vlse32.v v8, (a5), a3
; CHECK-NEXT:    addi a5, a0, 12
; CHECK-NEXT:    vlse32.v v9, (a5), a4
; CHECK-NEXT:    vadd.vv v8, v9, v8
; CHECK-NEXT:    vsse32.v v8, (a5), a4
; CHECK-NEXT:    addi a2, a2, -8
; CHECK-NEXT:    addi a1, a1, 512
; CHECK-NEXT:    addi a0, a0, 128
; CHECK-NEXT:    bnez a2, .LBB11_1
; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
; CHECK-NEXT:    ret
entry:
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.ind = phi <8 x i64> [ <i64 0, i64 4, i64 8, i64 12, i64 16, i64 20, i64 24, i64 28>, %entry ], [ %vec.ind.next, %vector.body ]
  %i = shl nuw nsw <8 x i64> %vec.ind, splat (i64 2)
  %i1 = getelementptr inbounds i32, ptr %B, <8 x i64> %i
  %wide.masked.gather = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i1, i32 4, <8 x i1> splat (i1 true), <8 x i32> undef)
  %i2 = getelementptr inbounds i32, ptr %A, <8 x i64> %vec.ind
  %wide.masked.gather52 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i2, i32 4, <8 x i1> splat (i1 true), <8 x i32> undef)
  %i3 = add nsw <8 x i32> %wide.masked.gather52, %wide.masked.gather
  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> %i3, <8 x ptr> %i2, i32 4, <8 x i1> splat (i1 true))
  %i4 = or disjoint <8 x i64> %vec.ind, splat (i64 1)
  %i5 = shl nsw <8 x i64> %i4, splat (i64 2)
  %i6 = getelementptr inbounds i32, ptr %B, <8 x i64> %i5
  %wide.masked.gather53 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i6, i32 4, <8 x i1> splat (i1 true), <8 x i32> undef)
  %i7 = getelementptr inbounds i32, ptr %A, <8 x i64> %i4
  %wide.masked.gather54 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i7, i32 4, <8 x i1> splat (i1 true), <8 x i32> undef)
  %i8 = add nsw <8 x i32> %wide.masked.gather54, %wide.masked.gather53
  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> %i8, <8 x ptr> %i7, i32 4, <8 x i1> splat (i1 true))
  %i9 = or disjoint <8 x i64> %vec.ind, splat (i64 2)
  %i10 = shl nsw <8 x i64> %i9, splat (i64 2)
  %i11 = getelementptr inbounds i32, ptr %B, <8 x i64> %i10
  %wide.masked.gather55 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i11, i32 4, <8 x i1> splat (i1 true), <8 x i32> undef)
  %i12 = getelementptr inbounds i32, ptr %A, <8 x i64> %i9
  %wide.masked.gather56 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i12, i32 4, <8 x i1> splat (i1 true), <8 x i32> undef)
  %i13 = add nsw <8 x i32> %wide.masked.gather56, %wide.masked.gather55
  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> %i13, <8 x ptr> %i12, i32 4, <8 x i1> splat (i1 true))
  %i14 = or disjoint <8 x i64> %vec.ind, <i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3, i64 3>
  %i15 = shl nsw <8 x i64> %i14, <i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2>
  %i16 = getelementptr inbounds i32, ptr %B, <8 x i64> %i15
  %wide.masked.gather57 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i16, i32 4, <8 x i1> splat (i1 true), <8 x i32> undef)
  %i17 = getelementptr inbounds i32, ptr %A, <8 x i64> %i14
  %wide.masked.gather58 = call <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr> %i17, i32 4, <8 x i1> splat (i1 true), <8 x i32> undef)
  %i18 = add nsw <8 x i32> %wide.masked.gather58, %wide.masked.gather57
  call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> %i18, <8 x ptr> %i17, i32 4, <8 x i1> splat (i1 true))
  %index.next = add nuw i64 %index, 8
  %vec.ind.next = add <8 x i64> %vec.ind, splat (i64 32)
  %i19 = icmp eq i64 %index.next, 256
  br i1 %i19, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}

declare <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr>, i32 immarg, <32 x i1>, <32 x i8>)
declare <8 x i32> @llvm.masked.gather.v8i32.v8p0(<8 x ptr>, i32 immarg, <8 x i1>, <8 x i32>)
declare void @llvm.masked.scatter.v32i8.v32p0(<32 x i8>, <32 x ptr>, i32 immarg, <32 x i1>)
declare void @llvm.masked.scatter.v8i32.v8p0(<8 x i32>, <8 x ptr>, i32 immarg, <8 x i1>)

; Make sure we don't crash in getTgtMemIntrinsic for a vector of pointers.
define void @gather_of_pointers(ptr noalias nocapture %arg, ptr noalias nocapture readonly %arg1) {
; V-LABEL: gather_of_pointers:
; V:       # %bb.0: # %bb
; V-NEXT:    lui a2, 2
; V-NEXT:    add a2, a0, a2
; V-NEXT:    li a3, 40
; V-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
; V-NEXT:  .LBB12_1: # %bb2
; V-NEXT:    # =>This Inner Loop Header: Depth=1
; V-NEXT:    addi a4, a1, 80
; V-NEXT:    vlse64.v v8, (a1), a3
; V-NEXT:    vlse64.v v9, (a4), a3
; V-NEXT:    addi a4, a0, 16
; V-NEXT:    vse64.v v8, (a0)
; V-NEXT:    vse64.v v9, (a4)
; V-NEXT:    addi a0, a0, 32
; V-NEXT:    addi a1, a1, 160
; V-NEXT:    bne a0, a2, .LBB12_1
; V-NEXT:  # %bb.2: # %bb18
; V-NEXT:    ret
;
; ZVE32F-LABEL: gather_of_pointers:
; ZVE32F:       # %bb.0: # %bb
; ZVE32F-NEXT:    li a2, 0
; ZVE32F-NEXT:    lui a3, 2
; ZVE32F-NEXT:    add a3, a0, a3
; ZVE32F-NEXT:    li a4, 1
; ZVE32F-NEXT:    li a5, 40
; ZVE32F-NEXT:  .LBB12_1: # %bb2
; ZVE32F-NEXT:    # =>This Inner Loop Header: Depth=1
; ZVE32F-NEXT:    mul a6, a4, a5
; ZVE32F-NEXT:    add a6, a1, a6
; ZVE32F-NEXT:    mul a7, a2, a5
; ZVE32F-NEXT:    add a7, a1, a7
; ZVE32F-NEXT:    ld t0, 0(a6)
; ZVE32F-NEXT:    ld t1, 0(a7)
; ZVE32F-NEXT:    ld a6, 80(a6)
; ZVE32F-NEXT:    ld a7, 80(a7)
; ZVE32F-NEXT:    sd t0, 8(a0)
; ZVE32F-NEXT:    sd t1, 0(a0)
; ZVE32F-NEXT:    sd a6, 24(a0)
; ZVE32F-NEXT:    sd a7, 16(a0)
; ZVE32F-NEXT:    addi a2, a2, 4
; ZVE32F-NEXT:    addi a0, a0, 32
; ZVE32F-NEXT:    addi a4, a4, 4
; ZVE32F-NEXT:    bne a0, a3, .LBB12_1
; ZVE32F-NEXT:  # %bb.2: # %bb18
; ZVE32F-NEXT:    ret
bb:
  br label %bb2

bb2:                                              ; preds = %bb2, %bb
  %i = phi i64 [ 0, %bb ], [ %i15, %bb2 ]
  %i3 = phi <2 x i64> [ <i64 0, i64 1>, %bb ], [ %i16, %bb2 ]
  %i4 = mul nuw nsw <2 x i64> %i3, splat (i64 5)
  %i5 = mul <2 x i64> %i3, splat (i64 5)
  %i6 = add <2 x i64> %i5, <i64 10, i64 10>
  %i7 = getelementptr inbounds ptr, ptr %arg1, <2 x i64> %i4
  %i8 = getelementptr inbounds ptr, ptr %arg1, <2 x i64> %i6
  %i9 = call <2 x ptr> @llvm.masked.gather.v2p0.v2p0(<2 x ptr> %i7, i32 8, <2 x i1> splat (i1 true), <2 x ptr> undef)
  %i10 = call <2 x ptr> @llvm.masked.gather.v2p0.v2p0(<2 x ptr> %i8, i32 8, <2 x i1> splat (i1 true), <2 x ptr> undef)
  %i11 = getelementptr inbounds ptr, ptr %arg, i64 %i
  store <2 x ptr> %i9, ptr %i11, align 8
  %i13 = getelementptr inbounds ptr, ptr %i11, i64 2
  store <2 x ptr> %i10, ptr %i13, align 8
  %i15 = add nuw i64 %i, 4
  %i16 = add <2 x i64> %i3, <i64 4, i64 4>
  %i17 = icmp eq i64 %i15, 1024
  br i1 %i17, label %bb18, label %bb2

bb18:                                             ; preds = %bb2
  ret void
}

declare <2 x ptr> @llvm.masked.gather.v2p0.v2p0(<2 x ptr>, i32 immarg, <2 x i1>, <2 x ptr>)

; Make sure we don't crash in getTgtMemIntrinsic for a vector of pointers.
define void @scatter_of_pointers(ptr noalias nocapture %arg, ptr noalias nocapture readonly %arg1) {
; V-LABEL: scatter_of_pointers:
; V:       # %bb.0: # %bb
; V-NEXT:    lui a2, 2
; V-NEXT:    add a2, a1, a2
; V-NEXT:    li a3, 40
; V-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
; V-NEXT:  .LBB13_1: # %bb2
; V-NEXT:    # =>This Inner Loop Header: Depth=1
; V-NEXT:    addi a4, a1, 16
; V-NEXT:    vle64.v v8, (a1)
; V-NEXT:    vle64.v v9, (a4)
; V-NEXT:    addi a4, a0, 80
; V-NEXT:    vsse64.v v8, (a0), a3
; V-NEXT:    vsse64.v v9, (a4), a3
; V-NEXT:    addi a1, a1, 32
; V-NEXT:    addi a0, a0, 160
; V-NEXT:    bne a1, a2, .LBB13_1
; V-NEXT:  # %bb.2: # %bb18
; V-NEXT:    ret
;
; ZVE32F-LABEL: scatter_of_pointers:
; ZVE32F:       # %bb.0: # %bb
; ZVE32F-NEXT:    li a2, 0
; ZVE32F-NEXT:    lui a3, 2
; ZVE32F-NEXT:    add a3, a1, a3
; ZVE32F-NEXT:    li a4, 1
; ZVE32F-NEXT:    li a5, 40
; ZVE32F-NEXT:  .LBB13_1: # %bb2
; ZVE32F-NEXT:    # =>This Inner Loop Header: Depth=1
; ZVE32F-NEXT:    ld a6, 0(a1)
; ZVE32F-NEXT:    ld a7, 8(a1)
; ZVE32F-NEXT:    ld t0, 16(a1)
; ZVE32F-NEXT:    ld t1, 24(a1)
; ZVE32F-NEXT:    mul t2, a4, a5
; ZVE32F-NEXT:    add t2, a0, t2
; ZVE32F-NEXT:    mul t3, a2, a5
; ZVE32F-NEXT:    add t3, a0, t3
; ZVE32F-NEXT:    sd a6, 0(t3)
; ZVE32F-NEXT:    sd a7, 0(t2)
; ZVE32F-NEXT:    sd t0, 80(t3)
; ZVE32F-NEXT:    sd t1, 80(t2)
; ZVE32F-NEXT:    addi a2, a2, 4
; ZVE32F-NEXT:    addi a1, a1, 32
; ZVE32F-NEXT:    addi a4, a4, 4
; ZVE32F-NEXT:    bne a1, a3, .LBB13_1
; ZVE32F-NEXT:  # %bb.2: # %bb18
; ZVE32F-NEXT:    ret
bb:
  br label %bb2

bb2:                                              ; preds = %bb2, %bb
  %i = phi i64 [ 0, %bb ], [ %i15, %bb2 ]
  %i3 = phi <2 x i64> [ <i64 0, i64 1>, %bb ], [ %i16, %bb2 ]
  %i4 = getelementptr inbounds ptr, ptr %arg1, i64 %i
  %i6 = load <2 x ptr>, ptr %i4, align 8
  %i7 = getelementptr inbounds ptr, ptr %i4, i64 2
  %i9 = load <2 x ptr>, ptr %i7, align 8
  %i10 = mul nuw nsw <2 x i64> %i3, splat (i64 5)
  %i11 = mul <2 x i64> %i3, splat (i64 5)
  %i12 = add <2 x i64> %i11, <i64 10, i64 10>
  %i13 = getelementptr inbounds ptr, ptr %arg, <2 x i64> %i10
  %i14 = getelementptr inbounds ptr, ptr %arg, <2 x i64> %i12
  call void @llvm.masked.scatter.v2p0.v2p0(<2 x ptr> %i6, <2 x ptr> %i13, i32 8, <2 x i1> splat (i1 true))
  call void @llvm.masked.scatter.v2p0.v2p0(<2 x ptr> %i9, <2 x ptr> %i14, i32 8, <2 x i1> splat (i1 true))
  %i15 = add nuw i64 %i, 4
  %i16 = add <2 x i64> %i3, <i64 4, i64 4>
  %i17 = icmp eq i64 %i15, 1024
  br i1 %i17, label %bb18, label %bb2

bb18:                                             ; preds = %bb2
  ret void
}

declare void @llvm.masked.scatter.v2p0.v2p0(<2 x ptr>, <2 x ptr>, i32 immarg, <2 x i1>)

define void @strided_load_startval_add_with_splat(ptr noalias nocapture %arg, ptr noalias nocapture readonly %arg1, i32 signext %arg2) {
; CHECK-LABEL: strided_load_startval_add_with_splat:
; CHECK:       # %bb.0: # %bb
; CHECK-NEXT:    li a3, 1024
; CHECK-NEXT:    beq a2, a3, .LBB14_7
; CHECK-NEXT:  # %bb.1: # %bb3
; CHECK-NEXT:    li a3, 1023
; CHECK-NEXT:    subw a5, a3, a2
; CHECK-NEXT:    li a6, 31
; CHECK-NEXT:    mv a4, a2
; CHECK-NEXT:    bltu a5, a6, .LBB14_5
; CHECK-NEXT:  # %bb.2: # %bb9
; CHECK-NEXT:    slli a5, a5, 32
; CHECK-NEXT:    srli a5, a5, 32
; CHECK-NEXT:    addi a5, a5, 1
; CHECK-NEXT:    andi a6, a5, -32
; CHECK-NEXT:    add a4, a6, a2
; CHECK-NEXT:    slli t0, a2, 2
; CHECK-NEXT:    add a7, a0, a2
; CHECK-NEXT:    add a2, a1, a2
; CHECK-NEXT:    add a2, a2, t0
; CHECK-NEXT:    add t0, a4, a0
; CHECK-NEXT:    li t2, 32
; CHECK-NEXT:    li t1, 5
; CHECK-NEXT:    vsetvli zero, t2, e8, m1, ta, ma
; CHECK-NEXT:  .LBB14_3: # %bb15
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    vlse8.v v8, (a2), t1
; CHECK-NEXT:    vle8.v v9, (a7)
; CHECK-NEXT:    vadd.vv v8, v9, v8
; CHECK-NEXT:    vse8.v v8, (a7)
; CHECK-NEXT:    addi a7, a7, 32
; CHECK-NEXT:    addi a2, a2, 160
; CHECK-NEXT:    bne a7, t0, .LBB14_3
; CHECK-NEXT:  # %bb.4: # %bb30
; CHECK-NEXT:    beq a5, a6, .LBB14_7
; CHECK-NEXT:  .LBB14_5: # %bb32
; CHECK-NEXT:    add a2, a0, a4
; CHECK-NEXT:    slli a5, a4, 2
; CHECK-NEXT:    add a1, a1, a4
; CHECK-NEXT:    add a1, a1, a5
; CHECK-NEXT:    subw a3, a3, a4
; CHECK-NEXT:    slli a3, a3, 32
; CHECK-NEXT:    srli a3, a3, 32
; CHECK-NEXT:    add a0, a4, a0
; CHECK-NEXT:    add a0, a0, a3
; CHECK-NEXT:    addi a0, a0, 1
; CHECK-NEXT:  .LBB14_6: # %bb35
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    lbu a3, 0(a1)
; CHECK-NEXT:    lbu a4, 0(a2)
; CHECK-NEXT:    add a3, a4, a3
; CHECK-NEXT:    sb a3, 0(a2)
; CHECK-NEXT:    addi a2, a2, 1
; CHECK-NEXT:    addi a1, a1, 5
; CHECK-NEXT:    bne a2, a0, .LBB14_6
; CHECK-NEXT:  .LBB14_7: # %bb34
; CHECK-NEXT:    ret
bb:
  %i = icmp eq i32 %arg2, 1024
  br i1 %i, label %bb34, label %bb3

bb3:                                              ; preds = %bb
  %i4 = sext i32 %arg2 to i64
  %i5 = sub i32 1023, %arg2
  %i6 = zext i32 %i5 to i64
  %i7 = add nuw nsw i64 %i6, 1
  %i8 = icmp ult i32 %i5, 31
  br i1 %i8, label %bb32, label %bb9

bb9:                                              ; preds = %bb3
  %i10 = and i64 %i7, 8589934560
  %i11 = add nsw i64 %i10, %i4
  %i12 = insertelement <32 x i64> poison, i64 %i4, i64 0
  %i13 = shufflevector <32 x i64> %i12, <32 x i64> poison, <32 x i32> zeroinitializer
  %i14 = add <32 x i64> %i13, <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15, i64 16, i64 17, i64 18, i64 19, i64 20, i64 21, i64 22, i64 23, i64 24, i64 25, i64 26, i64 27, i64 28, i64 29, i64 30, i64 31>
  br label %bb15

bb15:                                             ; preds = %bb15, %bb9
  %i16 = phi i64 [ 0, %bb9 ], [ %i27, %bb15 ]
  %i17 = phi <32 x i64> [ %i14, %bb9 ], [ %i28, %bb15 ]
  %i18 = add i64 %i16, %i4
  %i19 = mul nsw <32 x i64> %i17, splat (i64 5)
  %i20 = getelementptr inbounds i8, ptr %arg1, <32 x i64> %i19
  %i21 = call <32 x i8> @llvm.masked.gather.v32i8.v32p0(<32 x ptr> %i20, i32 1, <32 x i1> splat (i1 true), <32 x i8> undef)
  %i22 = getelementptr inbounds i8, ptr %arg, i64 %i18
  %i24 = load <32 x i8>, ptr %i22, align 1
  %i25 = add <32 x i8> %i24, %i21
  store <32 x i8> %i25, ptr %i22, align 1
  %i27 = add nuw i64 %i16, 32
  %i28 = add <32 x i64> %i17, splat (i64 32)
  %i29 = icmp eq i64 %i27, %i10
  br i1 %i29, label %bb30, label %bb15

bb30:                                             ; preds = %bb15
  %i31 = icmp eq i64 %i7, %i10
  br i1 %i31, label %bb34, label %bb32

bb32:                                             ; preds = %bb30, %bb3
  %i33 = phi i64 [ %i4, %bb3 ], [ %i11, %bb30 ]
  br label %bb35

bb34:                                             ; preds = %bb35, %bb30, %bb
  ret void

bb35:                                             ; preds = %bb35, %bb32
  %i36 = phi i64 [ %i43, %bb35 ], [ %i33, %bb32 ]
  %i37 = mul nsw i64 %i36, 5
  %i38 = getelementptr inbounds i8, ptr %arg1, i64 %i37
  %i39 = load i8, ptr %i38, align 1
  %i40 = getelementptr inbounds i8, ptr %arg, i64 %i36
  %i41 = load i8, ptr %i40, align 1
  %i42 = add i8 %i41, %i39
  store i8 %i42, ptr %i40, align 1
  %i43 = add nsw i64 %i36, 1
  %i44 = trunc i64 %i43 to i32
  %i45 = icmp eq i32 %i44, 1024
  br i1 %i45, label %bb34, label %bb35
}

declare <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr>, i32 immarg, <16 x i1>, <16 x i8>)
declare void @llvm.masked.scatter.v16i8.v16p0(<16 x i8>, <16 x ptr>, i32 immarg, <16 x i1>)

define void @gather_no_scalar_remainder(ptr noalias nocapture noundef %arg, ptr noalias nocapture noundef readonly %arg1, i64 noundef %arg2) {
; CHECK-LABEL: gather_no_scalar_remainder:
; CHECK:       # %bb.0: # %bb
; CHECK-NEXT:    slli a2, a2, 4
; CHECK-NEXT:    beqz a2, .LBB15_3
; CHECK-NEXT:  # %bb.1: # %bb2
; CHECK-NEXT:    addi a2, a2, -16
; CHECK-NEXT:    andi a2, a2, -16
; CHECK-NEXT:    add a2, a2, a0
; CHECK-NEXT:    addi a2, a2, 16
; CHECK-NEXT:    li a3, 5
; CHECK-NEXT:    vsetivli zero, 16, e8, mf2, ta, ma
; CHECK-NEXT:  .LBB15_2: # %bb4
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    vlse8.v v8, (a1), a3
; CHECK-NEXT:    vle8.v v9, (a0)
; CHECK-NEXT:    vadd.vv v8, v9, v8
; CHECK-NEXT:    vse8.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 16
; CHECK-NEXT:    addi a1, a1, 80
; CHECK-NEXT:    bne a0, a2, .LBB15_2
; CHECK-NEXT:  .LBB15_3: # %bb16
; CHECK-NEXT:    ret
bb:
  %i = shl i64 %arg2, 4
  %i3 = icmp eq i64 %i, 0
  br i1 %i3, label %bb16, label %bb2

bb2:                                              ; preds = %bb
  br label %bb4

bb4:                                              ; preds = %bb4, %bb2
  %i5 = phi i64 [ %i13, %bb4 ], [ 0, %bb2 ]
  %i6 = phi <16 x i64> [ %i14, %bb4 ], [ <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15>, %bb2 ]
  %i7 = mul <16 x i64> %i6, splat (i64 5)
  %i8 = getelementptr inbounds i8, ptr %arg1, <16 x i64> %i7
  %i9 = call <16 x i8> @llvm.masked.gather.v16i8.v16p0(<16 x ptr> %i8, i32 1, <16 x i1> splat (i1 true), <16 x i8> undef)
  %i10 = getelementptr inbounds i8, ptr %arg, i64 %i5
  %i11 = load <16 x i8>, ptr %i10, align 1
  %i12 = add <16 x i8> %i11, %i9
  store <16 x i8> %i12, ptr %i10, align 1
  %i13 = add nuw i64 %i5, 16
  %i14 = add <16 x i64> %i6, splat (i64 16)
  %i15 = icmp eq i64 %i13, %i
  br i1 %i15, label %bb16, label %bb4

bb16:                                             ; preds = %bb4, %bb
  ret void
}

define void @gather_zero_stride_fp(ptr noalias nocapture %A, ptr noalias nocapture readonly %B) {
; CHECK-LABEL: gather_zero_stride_fp:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    lui a2, 1
; CHECK-NEXT:    add a2, a0, a2
; CHECK-NEXT:    vsetivli zero, 8, e32, m1, ta, ma
; CHECK-NEXT:  .LBB16_1: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    flw fa5, 0(a1)
; CHECK-NEXT:    vle32.v v8, (a0)
; CHECK-NEXT:    vfadd.vf v8, v8, fa5
; CHECK-NEXT:    vse32.v v8, (a0)
; CHECK-NEXT:    addi a0, a0, 128
; CHECK-NEXT:    addi a1, a1, 640
; CHECK-NEXT:    bne a0, a2, .LBB16_1
; CHECK-NEXT:  # %bb.2: # %for.cond.cleanup
; CHECK-NEXT:    ret
entry:
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.ind = phi <8 x i64> [ zeroinitializer, %entry ], [ %vec.ind.next, %vector.body ]
  %i = mul nuw nsw <8 x i64> %vec.ind, splat (i64 5)
  %i1 = getelementptr inbounds float, ptr %B, <8 x i64> %i
  %wide.masked.gather = call <8 x float> @llvm.masked.gather.v8f32.v32p0(<8 x ptr> %i1, i32 4, <8 x i1> splat (i1 true), <8 x float> undef)
  %i2 = getelementptr inbounds float, ptr %A, i64 %index
  %wide.load = load <8 x float>, ptr %i2, align 4
  %i4 = fadd <8 x float> %wide.load, %wide.masked.gather
  store <8 x float> %i4, ptr %i2, align 4
  %index.next = add nuw i64 %index, 32
  %vec.ind.next = add <8 x i64> %vec.ind, splat (i64 32)
  %i6 = icmp eq i64 %index.next, 1024
  br i1 %i6, label %for.cond.cleanup, label %vector.body

for.cond.cleanup:                                 ; preds = %vector.body
  ret void
}