test/CodeGen/WebAssembly/interleaved-vectors.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc -mtriple=wasm32 -mattr=+simd128 %s -o - | FileCheck %s

target datalayout = "e-m:e-p:32:32-p10:8:8-p20:8:8-i64:64-i128:128-n32:64-S128-ni:1:10:20"
target triple = "wasm32"

define hidden void @i32_add(ptr noalias nofree noundef writeonly captures(none) %a, ptr nofree noundef readonly captures(none) %b, ptr nofree noundef readonly captures(none) %c, i32 noundef %N) {
; CHECK-LABEL: i32_add:
; CHECK:         .functype i32_add (i32, i32, i32, i32) -> ()
; CHECK-NEXT:    .local i32, i32, i32, i32, i32
; CHECK-NEXT:  # %bb.0: # %entry
; CHECK-NEXT:    block
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    i32.eqz
; CHECK-NEXT:    br_if 0 # 0: down to label0
; CHECK-NEXT:  # %bb.1: # %for.body.preheader
; CHECK-NEXT:    i32.const 0
; CHECK-NEXT:    local.set 4
; CHECK-NEXT:    block
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    i32.const 8
; CHECK-NEXT:    i32.lt_u
; CHECK-NEXT:    br_if 0 # 0: down to label1
; CHECK-NEXT:  # %bb.2: # %vector.ph
; CHECK-NEXT:    local.get 1
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    local.get 2
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 0
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    i32.const -8
; CHECK-NEXT:    i32.and
; CHECK-NEXT:    local.tee 4
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:  .LBB0_3: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    loop # label2:
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    i32x4.add
; CHECK-NEXT:    v128.store 0:p2align=2
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    i32x4.add
; CHECK-NEXT:    v128.store 0:p2align=2
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const -8
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.tee 8
; CHECK-NEXT:    br_if 0 # 0: up to label2
; CHECK-NEXT:  # %bb.4: # %middle.block
; CHECK-NEXT:    end_loop
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.eq
; CHECK-NEXT:    br_if 1 # 1: down to label0
; CHECK-NEXT:  .LBB0_5: # %for.body.preheader12
; CHECK-NEXT:    end_block # label1:
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.sub
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 1
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.const 2
; CHECK-NEXT:    i32.shl
; CHECK-NEXT:    local.tee 7
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    local.get 2
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 0
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:  .LBB0_6: # %for.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    loop # label3:
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.load 0
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.load 0
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    i32.store 0
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const -1
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.tee 8
; CHECK-NEXT:    br_if 0 # 0: up to label3
; CHECK-NEXT:  .LBB0_7: # %for.cond.cleanup
; CHECK-NEXT:    end_loop
; CHECK-NEXT:    end_block # label0:
; CHECK-NEXT:    # fallthrough-return
entry:
  %cmp7.not = icmp eq i32 %N, 0
  br i1 %cmp7.not, label %for.cond.cleanup, label %for.body.preheader

for.body.preheader:
  %min.iters.check = icmp ult i32 %N, 8
  br i1 %min.iters.check, label %for.body.preheader12, label %vector.ph

vector.ph:
  %n.vec = and i32 %N, -8
  br label %vector.body

vector.body:
  %index = phi i32 [ 0, %vector.ph ], [ %index.next, %vector.body ]
  %0 = getelementptr inbounds nuw [4 x i8], ptr %b, i32 %index
  %1 = getelementptr inbounds nuw i8, ptr %0, i32 16
  %wide.load = load <4 x i32>, ptr %0, align 4
  %wide.load9 = load <4 x i32>, ptr %1, align 4
  %2 = getelementptr inbounds nuw [4 x i8], ptr %c, i32 %index
  %3 = getelementptr inbounds nuw i8, ptr %2, i32 16
  %wide.load10 = load <4 x i32>, ptr %2, align 4
  %wide.load11 = load <4 x i32>, ptr %3, align 4
  %4 = add nsw <4 x i32> %wide.load10, %wide.load
  %5 = add nsw <4 x i32> %wide.load11, %wide.load9
  %6 = getelementptr inbounds nuw [4 x i8], ptr %a, i32 %index
  %7 = getelementptr inbounds nuw i8, ptr %6, i32 16
  store <4 x i32> %4, ptr %6, align 4
  store <4 x i32> %5, ptr %7, align 4
  %index.next = add nuw i32 %index, 8
  %8 = icmp eq i32 %index.next, %n.vec
  br i1 %8, label %middle.block, label %vector.body

middle.block:
  %cmp.n = icmp eq i32 %N, %n.vec
  br i1 %cmp.n, label %for.cond.cleanup, label %for.body.preheader12

for.body.preheader12:
  %i.08.ph = phi i32 [ 0, %for.body.preheader ], [ %n.vec, %middle.block ]
  br label %for.body

for.cond.cleanup:
  ret void

for.body:
  %i.08 = phi i32 [ %inc, %for.body ], [ %i.08.ph, %for.body.preheader12 ]
  %arrayidx = getelementptr inbounds nuw [4 x i8], ptr %b, i32 %i.08
  %9 = load i32, ptr %arrayidx, align 4
  %arrayidx1 = getelementptr inbounds nuw [4 x i8], ptr %c, i32 %i.08
  %10 = load i32, ptr %arrayidx1, align 4
  %add = add nsw i32 %10, %9
  %arrayidx2 = getelementptr inbounds nuw [4 x i8], ptr %a, i32 %i.08
  store i32 %add, ptr %arrayidx2, align 4
  %inc = add nuw i32 %i.08, 1
  %exitcond.not = icmp eq i32 %inc, %N
  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
}

define hidden void @i32_madd(ptr noalias nofree noundef writeonly captures(none) %a, ptr nofree noundef readonly captures(none) %b, ptr nofree noundef readonly captures(none) %c, ptr nofree noundef readonly captures(none) %d, i32 noundef %N) {
; CHECK-LABEL: i32_madd:
; CHECK:         .functype i32_madd (i32, i32, i32, i32, i32) -> ()
; CHECK-NEXT:    .local i32, i32, i32, i32, i32, i32
; CHECK-NEXT:  # %bb.0: # %entry
; CHECK-NEXT:    block
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.eqz
; CHECK-NEXT:    br_if 0 # 0: down to label4
; CHECK-NEXT:  # %bb.1: # %for.body.preheader
; CHECK-NEXT:    i32.const 0
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    block
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.const 8
; CHECK-NEXT:    i32.lt_u
; CHECK-NEXT:    br_if 0 # 0: down to label5
; CHECK-NEXT:  # %bb.2: # %vector.ph
; CHECK-NEXT:    local.get 1
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 2
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 0
; CHECK-NEXT:    local.set 9
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.const -8
; CHECK-NEXT:    i32.and
; CHECK-NEXT:    local.tee 5
; CHECK-NEXT:    local.set 10
; CHECK-NEXT:  .LBB1_3: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    loop # label6:
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    i32x4.mul
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    i32x4.add
; CHECK-NEXT:    v128.store 0:p2align=2
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    i32x4.mul
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    i32x4.add
; CHECK-NEXT:    v128.store 0:p2align=2
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 9
; CHECK-NEXT:    local.get 10
; CHECK-NEXT:    i32.const -8
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.tee 10
; CHECK-NEXT:    br_if 0 # 0: up to label6
; CHECK-NEXT:  # %bb.4: # %middle.block
; CHECK-NEXT:    end_loop
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.eq
; CHECK-NEXT:    br_if 1 # 1: down to label4
; CHECK-NEXT:  .LBB1_5: # %for.body.preheader16
; CHECK-NEXT:    end_block # label5:
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.sub
; CHECK-NEXT:    local.set 10
; CHECK-NEXT:    local.get 1
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.const 2
; CHECK-NEXT:    i32.shl
; CHECK-NEXT:    local.tee 9
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 2
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 0
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 9
; CHECK-NEXT:  .LBB1_6: # %for.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    loop # label7:
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.load 0
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.load 0
; CHECK-NEXT:    i32.mul
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.load 0
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    i32.store 0
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 9
; CHECK-NEXT:    local.get 10
; CHECK-NEXT:    i32.const -1
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.tee 10
; CHECK-NEXT:    br_if 0 # 0: up to label7
; CHECK-NEXT:  .LBB1_7: # %for.cond.cleanup
; CHECK-NEXT:    end_loop
; CHECK-NEXT:    end_block # label4:
; CHECK-NEXT:    # fallthrough-return
entry:
  %cmp9.not = icmp eq i32 %N, 0
  br i1 %cmp9.not, label %for.cond.cleanup, label %for.body.preheader

for.body.preheader:
  %min.iters.check = icmp ult i32 %N, 8
  br i1 %min.iters.check, label %for.body.preheader16, label %vector.ph

vector.ph:
  %n.vec = and i32 %N, -8
  br label %vector.body

vector.body:
  %index = phi i32 [ 0, %vector.ph ], [ %index.next, %vector.body ]
  %0 = getelementptr inbounds nuw [4 x i8], ptr %b, i32 %index
  %1 = getelementptr inbounds nuw i8, ptr %0, i32 16
  %wide.load = load <4 x i32>, ptr %0, align 4
  %wide.load11 = load <4 x i32>, ptr %1, align 4
  %2 = getelementptr inbounds nuw [4 x i8], ptr %c, i32 %index
  %3 = getelementptr inbounds nuw i8, ptr %2, i32 16
  %wide.load12 = load <4 x i32>, ptr %2, align 4
  %wide.load13 = load <4 x i32>, ptr %3, align 4
  %4 = mul nsw <4 x i32> %wide.load12, %wide.load
  %5 = mul nsw <4 x i32> %wide.load13, %wide.load11
  %6 = getelementptr inbounds nuw [4 x i8], ptr %d, i32 %index
  %7 = getelementptr inbounds nuw i8, ptr %6, i32 16
  %wide.load14 = load <4 x i32>, ptr %6, align 4
  %wide.load15 = load <4 x i32>, ptr %7, align 4
  %8 = add nsw <4 x i32> %4, %wide.load14
  %9 = add nsw <4 x i32> %5, %wide.load15
  %10 = getelementptr inbounds nuw [4 x i8], ptr %a, i32 %index
  %11 = getelementptr inbounds nuw i8, ptr %10, i32 16
  store <4 x i32> %8, ptr %10, align 4
  store <4 x i32> %9, ptr %11, align 4
  %index.next = add nuw i32 %index, 8
  %12 = icmp eq i32 %index.next, %n.vec
  br i1 %12, label %middle.block, label %vector.body

middle.block:
  %cmp.n = icmp eq i32 %N, %n.vec
  br i1 %cmp.n, label %for.cond.cleanup, label %for.body.preheader16

for.body.preheader16:
  %i.010.ph = phi i32 [ 0, %for.body.preheader ], [ %n.vec, %middle.block ]
  br label %for.body

for.cond.cleanup:
  ret void

for.body:
  %i.010 = phi i32 [ %inc, %for.body ], [ %i.010.ph, %for.body.preheader16 ]
  %arrayidx = getelementptr inbounds nuw [4 x i8], ptr %b, i32 %i.010
  %13 = load i32, ptr %arrayidx, align 4
  %arrayidx1 = getelementptr inbounds nuw [4 x i8], ptr %c, i32 %i.010
  %14 = load i32, ptr %arrayidx1, align 4
  %mul = mul nsw i32 %14, %13
  %arrayidx2 = getelementptr inbounds nuw [4 x i8], ptr %d, i32 %i.010
  %15 = load i32, ptr %arrayidx2, align 4
  %add = add nsw i32 %mul, %15
  %arrayidx3 = getelementptr inbounds nuw [4 x i8], ptr %a, i32 %i.010
  store i32 %add, ptr %arrayidx3, align 4
  %inc = add nuw i32 %i.010, 1
  %exitcond.not = icmp eq i32 %inc, %N
  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
}

define hidden void @i32_i16_add(ptr noalias nofree noundef writeonly captures(none) %a, ptr nofree noundef readonly captures(none) %b, ptr nofree noundef readonly captures(none) %c, i32 noundef %N) {
; CHECK-LABEL: i32_i16_add:
; CHECK:         .functype i32_i16_add (i32, i32, i32, i32) -> ()
; CHECK-NEXT:    .local i32, i32, i32, i32, i32
; CHECK-NEXT:  # %bb.0: # %entry
; CHECK-NEXT:    block
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    i32.eqz
; CHECK-NEXT:    br_if 0 # 0: down to label8
; CHECK-NEXT:  # %bb.1: # %for.body.preheader
; CHECK-NEXT:    i32.const 0
; CHECK-NEXT:    local.set 4
; CHECK-NEXT:    block
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    i32.const 8
; CHECK-NEXT:    i32.lt_u
; CHECK-NEXT:    br_if 0 # 0: down to label9
; CHECK-NEXT:  # %bb.2: # %vector.ph
; CHECK-NEXT:    local.get 1
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    local.get 2
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 0
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    i32.const -8
; CHECK-NEXT:    i32.and
; CHECK-NEXT:    local.tee 4
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:  .LBB2_3: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    loop # label10:
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32x4.load16x4_s 0:p2align=1
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32x4.load16x4_s 0:p2align=1
; CHECK-NEXT:    i32x4.add
; CHECK-NEXT:    v128.store 0:p2align=2
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 8
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    i32x4.load16x4_s 0:p2align=1
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.const 8
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    i32x4.load16x4_s 0:p2align=1
; CHECK-NEXT:    i32x4.add
; CHECK-NEXT:    v128.store 0:p2align=2
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const -8
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.tee 8
; CHECK-NEXT:    br_if 0 # 0: up to label10
; CHECK-NEXT:  # %bb.4: # %middle.block
; CHECK-NEXT:    end_loop
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.eq
; CHECK-NEXT:    br_if 1 # 1: down to label8
; CHECK-NEXT:  .LBB2_5: # %for.body.preheader13
; CHECK-NEXT:    end_block # label9:
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.sub
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 1
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.const 1
; CHECK-NEXT:    i32.shl
; CHECK-NEXT:    local.tee 6
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    local.get 2
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 0
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.const 2
; CHECK-NEXT:    i32.shl
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:  .LBB2_6: # %for.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    loop # label11:
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.load16_s 0
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.load16_s 0
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    i32.store 0
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.const 2
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 2
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const -1
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.tee 8
; CHECK-NEXT:    br_if 0 # 0: up to label11
; CHECK-NEXT:  .LBB2_7: # %for.cond.cleanup
; CHECK-NEXT:    end_loop
; CHECK-NEXT:    end_block # label8:
; CHECK-NEXT:    # fallthrough-return
entry:
  %cmp8.not = icmp eq i32 %N, 0
  br i1 %cmp8.not, label %for.cond.cleanup, label %for.body.preheader

for.body.preheader:
  %min.iters.check = icmp ult i32 %N, 8
  br i1 %min.iters.check, label %for.body.preheader13, label %vector.ph

vector.ph:
  %n.vec = and i32 %N, -8
  br label %vector.body

vector.body:
  %index = phi i32 [ 0, %vector.ph ], [ %index.next, %vector.body ]
  %0 = getelementptr inbounds nuw [2 x i8], ptr %b, i32 %index
  %1 = getelementptr inbounds nuw i8, ptr %0, i32 8
  %wide.load = load <4 x i16>, ptr %0, align 2
  %wide.load10 = load <4 x i16>, ptr %1, align 2
  %2 = sext <4 x i16> %wide.load to <4 x i32>
  %3 = sext <4 x i16> %wide.load10 to <4 x i32>
  %4 = getelementptr inbounds nuw [2 x i8], ptr %c, i32 %index
  %5 = getelementptr inbounds nuw i8, ptr %4, i32 8
  %wide.load11 = load <4 x i16>, ptr %4, align 2
  %wide.load12 = load <4 x i16>, ptr %5, align 2
  %6 = sext <4 x i16> %wide.load11 to <4 x i32>
  %7 = sext <4 x i16> %wide.load12 to <4 x i32>
  %8 = add nsw <4 x i32> %6, %2
  %9 = add nsw <4 x i32> %7, %3
  %10 = getelementptr inbounds nuw [4 x i8], ptr %a, i32 %index
  %11 = getelementptr inbounds nuw i8, ptr %10, i32 16
  store <4 x i32> %8, ptr %10, align 4
  store <4 x i32> %9, ptr %11, align 4
  %index.next = add nuw i32 %index, 8
  %12 = icmp eq i32 %index.next, %n.vec
  br i1 %12, label %middle.block, label %vector.body

middle.block:
  %cmp.n = icmp eq i32 %N, %n.vec
  br i1 %cmp.n, label %for.cond.cleanup, label %for.body.preheader13

for.body.preheader13:
  %i.09.ph = phi i32 [ 0, %for.body.preheader ], [ %n.vec, %middle.block ]
  br label %for.body

for.cond.cleanup:
  ret void

for.body:
  %i.09 = phi i32 [ %inc, %for.body ], [ %i.09.ph, %for.body.preheader13 ]
  %arrayidx = getelementptr inbounds nuw [2 x i8], ptr %b, i32 %i.09
  %13 = load i16, ptr %arrayidx, align 2
  %conv = sext i16 %13 to i32
  %arrayidx1 = getelementptr inbounds nuw [2 x i8], ptr %c, i32 %i.09
  %14 = load i16, ptr %arrayidx1, align 2
  %conv2 = sext i16 %14 to i32
  %add = add nsw i32 %conv2, %conv
  %arrayidx3 = getelementptr inbounds nuw [4 x i8], ptr %a, i32 %i.09
  store i32 %add, ptr %arrayidx3, align 4
  %inc = add nuw i32 %i.09, 1
  %exitcond.not = icmp eq i32 %inc, %N
  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
}

define hidden void @i32_i8_madd(ptr noalias nofree noundef writeonly captures(none) %a, ptr nofree noundef readonly captures(none) %b, ptr nofree noundef readonly captures(none) %c, ptr nofree noundef readonly captures(none) %d, i32 noundef %N) {
; CHECK-LABEL: i32_i8_madd:
; CHECK:         .functype i32_i8_madd (i32, i32, i32, i32, i32) -> ()
; CHECK-NEXT:    .local i32, i32, i32, i32, i32, i32
; CHECK-NEXT:  # %bb.0: # %entry
; CHECK-NEXT:    block
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.eqz
; CHECK-NEXT:    br_if 0 # 0: down to label12
; CHECK-NEXT:  # %bb.1: # %for.body.preheader
; CHECK-NEXT:    i32.const 0
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    block
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.const 8
; CHECK-NEXT:    i32.lt_u
; CHECK-NEXT:    br_if 0 # 0: down to label13
; CHECK-NEXT:  # %bb.2: # %vector.ph
; CHECK-NEXT:    local.get 1
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 2
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 0
; CHECK-NEXT:    local.set 9
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.const -8
; CHECK-NEXT:    i32.and
; CHECK-NEXT:    local.tee 5
; CHECK-NEXT:    local.set 10
; CHECK-NEXT:  .LBB3_3: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    loop # label14:
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    v128.load32_zero 0:p2align=0
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    v128.load32_zero 0:p2align=0
; CHECK-NEXT:    i16x8.extmul_low_i8x16_s
; CHECK-NEXT:    i32x4.extend_low_i16x8_s
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    v128.load32_zero 0:p2align=0
; CHECK-NEXT:    i16x8.extend_low_i8x16_s
; CHECK-NEXT:    i32x4.extend_low_i16x8_s
; CHECK-NEXT:    i32x4.add
; CHECK-NEXT:    v128.store 0:p2align=2
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    v128.load32_zero 0:p2align=0
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    v128.load32_zero 0:p2align=0
; CHECK-NEXT:    i16x8.extmul_low_i8x16_s
; CHECK-NEXT:    i32x4.extend_low_i16x8_s
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    v128.load32_zero 0:p2align=0
; CHECK-NEXT:    i16x8.extend_low_i8x16_s
; CHECK-NEXT:    i32x4.extend_low_i16x8_s
; CHECK-NEXT:    i32x4.add
; CHECK-NEXT:    v128.store 0:p2align=2
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 8
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 8
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const 8
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 9
; CHECK-NEXT:    local.get 10
; CHECK-NEXT:    i32.const -8
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.tee 10
; CHECK-NEXT:    br_if 0 # 0: up to label14
; CHECK-NEXT:  # %bb.4: # %middle.block
; CHECK-NEXT:    end_loop
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.eq
; CHECK-NEXT:    br_if 1 # 1: down to label12
; CHECK-NEXT:  .LBB3_5: # %for.body.preheader18
; CHECK-NEXT:    end_block # label13:
; CHECK-NEXT:    local.get 1
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 2
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.sub
; CHECK-NEXT:    local.set 10
; CHECK-NEXT:    local.get 0
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.const 2
; CHECK-NEXT:    i32.shl
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 9
; CHECK-NEXT:  .LBB3_6: # %for.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    loop # label15:
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.load8_s 0
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.load8_s 0
; CHECK-NEXT:    i32.mul
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.load8_s 0
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    i32.store 0
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 1
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 1
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const 1
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 9
; CHECK-NEXT:    local.get 10
; CHECK-NEXT:    i32.const -1
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.tee 10
; CHECK-NEXT:    br_if 0 # 0: up to label15
; CHECK-NEXT:  .LBB3_7: # %for.cond.cleanup
; CHECK-NEXT:    end_loop
; CHECK-NEXT:    end_block # label12:
; CHECK-NEXT:    # fallthrough-return
entry:
  %cmp11.not = icmp eq i32 %N, 0
  br i1 %cmp11.not, label %for.cond.cleanup, label %for.body.preheader

for.body.preheader:
  %min.iters.check = icmp ult i32 %N, 8
  br i1 %min.iters.check, label %for.body.preheader18, label %vector.ph

vector.ph:
  %n.vec = and i32 %N, -8
  br label %vector.body

vector.body:
  %index = phi i32 [ 0, %vector.ph ], [ %index.next, %vector.body ]
  %0 = getelementptr inbounds nuw i8, ptr %b, i32 %index
  %1 = getelementptr inbounds nuw i8, ptr %0, i32 4
  %wide.load = load <4 x i8>, ptr %0, align 1
  %wide.load13 = load <4 x i8>, ptr %1, align 1
  %2 = sext <4 x i8> %wide.load to <4 x i32>
  %3 = sext <4 x i8> %wide.load13 to <4 x i32>
  %4 = getelementptr inbounds nuw i8, ptr %c, i32 %index
  %5 = getelementptr inbounds nuw i8, ptr %4, i32 4
  %wide.load14 = load <4 x i8>, ptr %4, align 1
  %wide.load15 = load <4 x i8>, ptr %5, align 1
  %6 = sext <4 x i8> %wide.load14 to <4 x i32>
  %7 = sext <4 x i8> %wide.load15 to <4 x i32>
  %8 = mul nsw <4 x i32> %6, %2
  %9 = mul nsw <4 x i32> %7, %3
  %10 = getelementptr inbounds nuw i8, ptr %d, i32 %index
  %11 = getelementptr inbounds nuw i8, ptr %10, i32 4
  %wide.load16 = load <4 x i8>, ptr %10, align 1
  %wide.load17 = load <4 x i8>, ptr %11, align 1
  %12 = sext <4 x i8> %wide.load16 to <4 x i32>
  %13 = sext <4 x i8> %wide.load17 to <4 x i32>
  %14 = add nsw <4 x i32> %8, %12
  %15 = add nsw <4 x i32> %9, %13
  %16 = getelementptr inbounds nuw [4 x i8], ptr %a, i32 %index
  %17 = getelementptr inbounds nuw i8, ptr %16, i32 16
  store <4 x i32> %14, ptr %16, align 4
  store <4 x i32> %15, ptr %17, align 4
  %index.next = add nuw i32 %index, 8
  %18 = icmp eq i32 %index.next, %n.vec
  br i1 %18, label %middle.block, label %vector.body

middle.block:
  %cmp.n = icmp eq i32 %N, %n.vec
  br i1 %cmp.n, label %for.cond.cleanup, label %for.body.preheader18

for.body.preheader18:
  %i.012.ph = phi i32 [ 0, %for.body.preheader ], [ %n.vec, %middle.block ]
  br label %for.body

for.cond.cleanup:
  ret void

for.body:
  %i.012 = phi i32 [ %inc, %for.body ], [ %i.012.ph, %for.body.preheader18 ]
  %arrayidx = getelementptr inbounds nuw i8, ptr %b, i32 %i.012
  %19 = load i8, ptr %arrayidx, align 1
  %conv = sext i8 %19 to i32
  %arrayidx1 = getelementptr inbounds nuw i8, ptr %c, i32 %i.012
  %20 = load i8, ptr %arrayidx1, align 1
  %conv2 = sext i8 %20 to i32
  %mul = mul nsw i32 %conv2, %conv
  %arrayidx3 = getelementptr inbounds nuw i8, ptr %d, i32 %i.012
  %21 = load i8, ptr %arrayidx3, align 1
  %conv4 = sext i8 %21 to i32
  %add = add nsw i32 %mul, %conv4
  %arrayidx5 = getelementptr inbounds nuw [4 x i8], ptr %a, i32 %i.012
  store i32 %add, ptr %arrayidx5, align 4
  %inc = add nuw i32 %i.012, 1
  %exitcond.not = icmp eq i32 %inc, %N
  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
}

define hidden void @f32_add(ptr noalias nofree noundef writeonly captures(none) %a, ptr nofree noundef readonly captures(none) %b, ptr nofree noundef readonly captures(none) %c, i32 noundef %N) {
; CHECK-LABEL: f32_add:
; CHECK:         .functype f32_add (i32, i32, i32, i32) -> ()
; CHECK-NEXT:    .local i32, i32, i32, i32, i32
; CHECK-NEXT:  # %bb.0: # %entry
; CHECK-NEXT:    block
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    i32.eqz
; CHECK-NEXT:    br_if 0 # 0: down to label16
; CHECK-NEXT:  # %bb.1: # %for.body.preheader
; CHECK-NEXT:    i32.const 0
; CHECK-NEXT:    local.set 4
; CHECK-NEXT:    block
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    i32.const 8
; CHECK-NEXT:    i32.lt_u
; CHECK-NEXT:    br_if 0 # 0: down to label17
; CHECK-NEXT:  # %bb.2: # %vector.ph
; CHECK-NEXT:    local.get 1
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    local.get 2
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 0
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    i32.const -8
; CHECK-NEXT:    i32.and
; CHECK-NEXT:    local.tee 4
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:  .LBB4_3: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    loop # label18:
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    f32x4.add
; CHECK-NEXT:    v128.store 0:p2align=2
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    f32x4.add
; CHECK-NEXT:    v128.store 0:p2align=2
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const -8
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.tee 8
; CHECK-NEXT:    br_if 0 # 0: up to label18
; CHECK-NEXT:  # %bb.4: # %middle.block
; CHECK-NEXT:    end_loop
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.eq
; CHECK-NEXT:    br_if 1 # 1: down to label16
; CHECK-NEXT:  .LBB4_5: # %for.body.preheader12
; CHECK-NEXT:    end_block # label17:
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.sub
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 1
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.const 2
; CHECK-NEXT:    i32.shl
; CHECK-NEXT:    local.tee 7
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    local.get 2
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 0
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:  .LBB4_6: # %for.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    loop # label19:
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    f32.load 0
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    f32.load 0
; CHECK-NEXT:    f32.add
; CHECK-NEXT:    f32.store 0
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const -1
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.tee 8
; CHECK-NEXT:    br_if 0 # 0: up to label19
; CHECK-NEXT:  .LBB4_7: # %for.cond.cleanup
; CHECK-NEXT:    end_loop
; CHECK-NEXT:    end_block # label16:
; CHECK-NEXT:    # fallthrough-return
entry:
  %cmp7.not = icmp eq i32 %N, 0
  br i1 %cmp7.not, label %for.cond.cleanup, label %for.body.preheader

for.body.preheader:
  %min.iters.check = icmp ult i32 %N, 8
  br i1 %min.iters.check, label %for.body.preheader12, label %vector.ph

vector.ph:
  %n.vec = and i32 %N, -8
  br label %vector.body

vector.body:
  %index = phi i32 [ 0, %vector.ph ], [ %index.next, %vector.body ]
  %0 = getelementptr inbounds nuw [4 x i8], ptr %b, i32 %index
  %1 = getelementptr inbounds nuw i8, ptr %0, i32 16
  %wide.load = load <4 x float>, ptr %0, align 4
  %wide.load9 = load <4 x float>, ptr %1, align 4
  %2 = getelementptr inbounds nuw [4 x i8], ptr %c, i32 %index
  %3 = getelementptr inbounds nuw i8, ptr %2, i32 16
  %wide.load10 = load <4 x float>, ptr %2, align 4
  %wide.load11 = load <4 x float>, ptr %3, align 4
  %4 = fadd <4 x float> %wide.load, %wide.load10
  %5 = fadd <4 x float> %wide.load9, %wide.load11
  %6 = getelementptr inbounds nuw [4 x i8], ptr %a, i32 %index
  %7 = getelementptr inbounds nuw i8, ptr %6, i32 16
  store <4 x float> %4, ptr %6, align 4
  store <4 x float> %5, ptr %7, align 4
  %index.next = add nuw i32 %index, 8
  %8 = icmp eq i32 %index.next, %n.vec
  br i1 %8, label %middle.block, label %vector.body

middle.block:
  %cmp.n = icmp eq i32 %N, %n.vec
  br i1 %cmp.n, label %for.cond.cleanup, label %for.body.preheader12

for.body.preheader12:
  %i.08.ph = phi i32 [ 0, %for.body.preheader ], [ %n.vec, %middle.block ]
  br label %for.body

for.cond.cleanup:
  ret void

for.body:
  %i.08 = phi i32 [ %inc, %for.body ], [ %i.08.ph, %for.body.preheader12 ]
  %arrayidx = getelementptr inbounds nuw [4 x i8], ptr %b, i32 %i.08
  %9 = load float, ptr %arrayidx, align 4
  %arrayidx1 = getelementptr inbounds nuw [4 x i8], ptr %c, i32 %i.08
  %10 = load float, ptr %arrayidx1, align 4
  %add = fadd float %9, %10
  %arrayidx2 = getelementptr inbounds nuw [4 x i8], ptr %a, i32 %i.08
  store float %add, ptr %arrayidx2, align 4
  %inc = add nuw i32 %i.08, 1
  %exitcond.not = icmp eq i32 %inc, %N
  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
}

define hidden void @f32_madd(ptr noalias nofree noundef writeonly captures(none) %a, ptr nofree noundef readonly captures(none) %b, ptr nofree noundef readonly captures(none) %c, ptr nofree noundef readonly captures(none) %d, i32 noundef %N) {
; CHECK-LABEL: f32_madd:
; CHECK:         .functype f32_madd (i32, i32, i32, i32, i32) -> ()
; CHECK-NEXT:    .local i32, i32, i32, i32, i32, i32
; CHECK-NEXT:  # %bb.0: # %entry
; CHECK-NEXT:    block
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.eqz
; CHECK-NEXT:    br_if 0 # 0: down to label20
; CHECK-NEXT:  # %bb.1: # %for.body.preheader
; CHECK-NEXT:    i32.const 0
; CHECK-NEXT:    local.set 5
; CHECK-NEXT:    block
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.const 8
; CHECK-NEXT:    i32.lt_u
; CHECK-NEXT:    br_if 0 # 0: down to label21
; CHECK-NEXT:  # %bb.2: # %vector.ph
; CHECK-NEXT:    local.get 1
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 2
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 0
; CHECK-NEXT:    local.set 9
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    i32.const -8
; CHECK-NEXT:    i32.and
; CHECK-NEXT:    local.tee 5
; CHECK-NEXT:    local.set 10
; CHECK-NEXT:  .LBB5_3: # %vector.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    loop # label22:
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    f32x4.mul
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    f32x4.add
; CHECK-NEXT:    v128.store 0:p2align=2
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    f32x4.mul
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const 16
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    v128.load 0:p2align=2
; CHECK-NEXT:    f32x4.add
; CHECK-NEXT:    v128.store 0:p2align=2
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.const 32
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 9
; CHECK-NEXT:    local.get 10
; CHECK-NEXT:    i32.const -8
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.tee 10
; CHECK-NEXT:    br_if 0 # 0: up to label22
; CHECK-NEXT:  # %bb.4: # %middle.block
; CHECK-NEXT:    end_loop
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.eq
; CHECK-NEXT:    br_if 1 # 1: down to label20
; CHECK-NEXT:  .LBB5_5: # %for.body.preheader16
; CHECK-NEXT:    end_block # label21:
; CHECK-NEXT:    local.get 4
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.sub
; CHECK-NEXT:    local.set 10
; CHECK-NEXT:    local.get 1
; CHECK-NEXT:    local.get 5
; CHECK-NEXT:    i32.const 2
; CHECK-NEXT:    i32.shl
; CHECK-NEXT:    local.tee 9
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 2
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 3
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 0
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 9
; CHECK-NEXT:  .LBB5_6: # %for.body
; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    loop # label23:
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    f32.load 0
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    f32.load 0
; CHECK-NEXT:    f32.mul
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    f32.load 0
; CHECK-NEXT:    f32.add
; CHECK-NEXT:    f32.store 0
; CHECK-NEXT:    local.get 6
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 6
; CHECK-NEXT:    local.get 7
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 7
; CHECK-NEXT:    local.get 8
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 8
; CHECK-NEXT:    local.get 9
; CHECK-NEXT:    i32.const 4
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.set 9
; CHECK-NEXT:    local.get 10
; CHECK-NEXT:    i32.const -1
; CHECK-NEXT:    i32.add
; CHECK-NEXT:    local.tee 10
; CHECK-NEXT:    br_if 0 # 0: up to label23
; CHECK-NEXT:  .LBB5_7: # %for.cond.cleanup
; CHECK-NEXT:    end_loop
; CHECK-NEXT:    end_block # label20:
; CHECK-NEXT:    # fallthrough-return
entry:
  %cmp9.not = icmp eq i32 %N, 0
  br i1 %cmp9.not, label %for.cond.cleanup, label %for.body.preheader

for.body.preheader:
  %min.iters.check = icmp ult i32 %N, 8
  br i1 %min.iters.check, label %for.body.preheader16, label %vector.ph

vector.ph:
  %n.vec = and i32 %N, -8
  br label %vector.body

vector.body:
  %index = phi i32 [ 0, %vector.ph ], [ %index.next, %vector.body ]
  %0 = getelementptr inbounds nuw [4 x i8], ptr %b, i32 %index
  %1 = getelementptr inbounds nuw i8, ptr %0, i32 16
  %wide.load = load <4 x float>, ptr %0, align 4
  %wide.load11 = load <4 x float>, ptr %1, align 4
  %2 = getelementptr inbounds nuw [4 x i8], ptr %c, i32 %index
  %3 = getelementptr inbounds nuw i8, ptr %2, i32 16
  %wide.load12 = load <4 x float>, ptr %2, align 4
  %wide.load13 = load <4 x float>, ptr %3, align 4
  %4 = getelementptr inbounds nuw [4 x i8], ptr %d, i32 %index
  %5 = getelementptr inbounds nuw i8, ptr %4, i32 16
  %wide.load14 = load <4 x float>, ptr %4, align 4
  %wide.load15 = load <4 x float>, ptr %5, align 4
  %6 = tail call <4 x float> @llvm.fmuladd.v4f32(<4 x float> %wide.load, <4 x float> %wide.load12, <4 x float> %wide.load14)
  %7 = tail call <4 x float> @llvm.fmuladd.v4f32(<4 x float> %wide.load11, <4 x float> %wide.load13, <4 x float> %wide.load15)
  %8 = getelementptr inbounds nuw [4 x i8], ptr %a, i32 %index
  %9 = getelementptr inbounds nuw i8, ptr %8, i32 16
  store <4 x float> %6, ptr %8, align 4
  store <4 x float> %7, ptr %9, align 4
  %index.next = add nuw i32 %index, 8
  %10 = icmp eq i32 %index.next, %n.vec
  br i1 %10, label %middle.block, label %vector.body

middle.block:
  %cmp.n = icmp eq i32 %N, %n.vec
  br i1 %cmp.n, label %for.cond.cleanup, label %for.body.preheader16

for.body.preheader16:
  %i.010.ph = phi i32 [ 0, %for.body.preheader ], [ %n.vec, %middle.block ]
  br label %for.body

for.cond.cleanup:
  ret void

for.body:
  %i.010 = phi i32 [ %inc, %for.body ], [ %i.010.ph, %for.body.preheader16 ]
  %arrayidx = getelementptr inbounds nuw [4 x i8], ptr %b, i32 %i.010
  %11 = load float, ptr %arrayidx, align 4
  %arrayidx1 = getelementptr inbounds nuw [4 x i8], ptr %c, i32 %i.010
  %12 = load float, ptr %arrayidx1, align 4
  %arrayidx2 = getelementptr inbounds nuw [4 x i8], ptr %d, i32 %i.010
  %13 = load float, ptr %arrayidx2, align 4
  %14 = tail call float @llvm.fmuladd.f32(float %11, float %12, float %13)
  %arrayidx3 = getelementptr inbounds nuw [4 x i8], ptr %a, i32 %i.010
  store float %14, ptr %arrayidx3, align 4
  %inc = add nuw i32 %i.010, 1
  %exitcond.not = icmp eq i32 %inc, %N
  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
}

declare float @llvm.fmuladd.f32(float, float, float)

declare <4 x float> @llvm.fmuladd.v4f32(<4 x float>, <4 x float>, <4 x float>)