path: root/llvm/test/CodeGen/X86/avg.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+sse2 | FileCheck %s --check-prefix=SSE2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+avx2 | FileCheck %s --check-prefix=AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+avx512f | FileCheck %s --check-prefix=AVX512F
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+avx512bw | FileCheck %s --check-prefix=AVX512BW

define void @avg_v4i8(<4 x i8>* %a, <4 x i8>* %b) {
; SSE2-LABEL: avg_v4i8:
; SSE2:       # BB#0:
; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; SSE2-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
; SSE2-NEXT:    pavgb %xmm0, %xmm1
; SSE2-NEXT:    movd %xmm1, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v4i8:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX2-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
; AVX2-NEXT:    vpavgb %xmm0, %xmm1, %xmm0
; AVX2-NEXT:    vmovd %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v4i8:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX512F-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
; AVX512F-NEXT:    vpavgb %xmm0, %xmm1, %xmm0
; AVX512F-NEXT:    vmovd %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v4i8:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX512BW-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
; AVX512BW-NEXT:    vpavgb %xmm0, %xmm1, %xmm0
; AVX512BW-NEXT:    vmovd %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <4 x i8>, <4 x i8>* %a
  %2 = load <4 x i8>, <4 x i8>* %b
  %3 = zext <4 x i8> %1 to <4 x i32>
  %4 = zext <4 x i8> %2 to <4 x i32>
  %5 = add nuw nsw <4 x i32> %3, <i32 1, i32 1, i32 1, i32 1>
  %6 = add nuw nsw <4 x i32> %5, %4
  %7 = lshr <4 x i32> %6, <i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <4 x i32> %7 to <4 x i8>
  store <4 x i8> %8, <4 x i8>* undef, align 4
  ret void
}

define void @avg_v8i8(<8 x i8>* %a, <8 x i8>* %b) {
; SSE2-LABEL: avg_v8i8:
; SSE2:       # BB#0:
; SSE2-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
; SSE2-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
; SSE2-NEXT:    pavgb %xmm0, %xmm1
; SSE2-NEXT:    movq %xmm1, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v8i8:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
; AVX2-NEXT:    vpavgb %xmm0, %xmm1, %xmm0
; AVX2-NEXT:    vmovq %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v8i8:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512F-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
; AVX512F-NEXT:    vpavgb %xmm0, %xmm1, %xmm0
; AVX512F-NEXT:    vmovq %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v8i8:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512BW-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
; AVX512BW-NEXT:    vpavgb %xmm0, %xmm1, %xmm0
; AVX512BW-NEXT:    vmovq %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <8 x i8>, <8 x i8>* %a
  %2 = load <8 x i8>, <8 x i8>* %b
  %3 = zext <8 x i8> %1 to <8 x i32>
  %4 = zext <8 x i8> %2 to <8 x i32>
  %5 = add nuw nsw <8 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %6 = add nuw nsw <8 x i32> %5, %4
  %7 = lshr <8 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <8 x i32> %7 to <8 x i8>
  store <8 x i8> %8, <8 x i8>* undef, align 4
  ret void
}

define void @avg_v16i8(<16 x i8>* %a, <16 x i8>* %b) {
; SSE2-LABEL: avg_v16i8:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rsi), %xmm0
; SSE2-NEXT:    pavgb (%rdi), %xmm0
; SSE2-NEXT:    movdqu %xmm0, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v16i8:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovdqa (%rsi), %xmm0
; AVX2-NEXT:    vpavgb (%rdi), %xmm0, %xmm0
; AVX2-NEXT:    vmovdqu %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v16i8:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rsi), %xmm0
; AVX512F-NEXT:    vpavgb (%rdi), %xmm0, %xmm0
; AVX512F-NEXT:    vmovdqu %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v16i8:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rsi), %xmm0
; AVX512BW-NEXT:    vpavgb (%rdi), %xmm0, %xmm0
; AVX512BW-NEXT:    vmovdqu %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <16 x i8>, <16 x i8>* %a
  %2 = load <16 x i8>, <16 x i8>* %b
  %3 = zext <16 x i8> %1 to <16 x i32>
  %4 = zext <16 x i8> %2 to <16 x i32>
  %5 = add nuw nsw <16 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %6 = add nuw nsw <16 x i32> %5, %4
  %7 = lshr <16 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <16 x i32> %7 to <16 x i8>
  store <16 x i8> %8, <16 x i8>* undef, align 4
  ret void
}

define void @avg_v32i8(<32 x i8>* %a, <32 x i8>* %b) {
; SSE2-LABEL: avg_v32i8:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm8
; SSE2-NEXT:    movdqa 16(%rdi), %xmm11
; SSE2-NEXT:    movdqa (%rsi), %xmm0
; SSE2-NEXT:    movdqa 16(%rsi), %xmm1
; SSE2-NEXT:    pxor %xmm4, %xmm4
; SSE2-NEXT:    pshufd {{.*#+}} xmm10 = xmm8[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm8 = xmm8[0],xmm4[0],xmm8[1],xmm4[1],xmm8[2],xmm4[2],xmm8[3],xmm4[3],xmm8[4],xmm4[4],xmm8[5],xmm4[5],xmm8[6],xmm4[6],xmm8[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm8, %xmm2
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm4[4],xmm2[5],xmm4[5],xmm2[6],xmm4[6],xmm2[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm2, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm4[0],xmm8[1],xmm4[1],xmm8[2],xmm4[2],xmm8[3],xmm4[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm10 = xmm10[0],xmm4[0],xmm10[1],xmm4[1],xmm10[2],xmm4[2],xmm10[3],xmm4[3],xmm10[4],xmm4[4],xmm10[5],xmm4[5],xmm10[6],xmm4[6],xmm10[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm10, %xmm12
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm12 = xmm12[4],xmm4[4],xmm12[5],xmm4[5],xmm12[6],xmm4[6],xmm12[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm10 = xmm10[0],xmm4[0],xmm10[1],xmm4[1],xmm10[2],xmm4[2],xmm10[3],xmm4[3]
; SSE2-NEXT:    pshufd {{.*#+}} xmm15 = xmm11[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm11 = xmm11[0],xmm4[0],xmm11[1],xmm4[1],xmm11[2],xmm4[2],xmm11[3],xmm4[3],xmm11[4],xmm4[4],xmm11[5],xmm4[5],xmm11[6],xmm4[6],xmm11[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm11, %xmm14
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm14 = xmm14[4],xmm4[4],xmm14[5],xmm4[5],xmm14[6],xmm4[6],xmm14[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm11 = xmm11[0],xmm4[0],xmm11[1],xmm4[1],xmm11[2],xmm4[2],xmm11[3],xmm4[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm4[0],xmm15[1],xmm4[1],xmm15[2],xmm4[2],xmm15[3],xmm4[3],xmm15[4],xmm4[4],xmm15[5],xmm4[5],xmm15[6],xmm4[6],xmm15[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm15, %xmm9
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm9 = xmm9[4],xmm4[4],xmm9[5],xmm4[5],xmm9[6],xmm4[6],xmm9[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm15 = xmm15[0],xmm4[0],xmm15[1],xmm4[1],xmm15[2],xmm4[2],xmm15[3],xmm4[3]
; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm0, %xmm7
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm7 = xmm7[4],xmm4[4],xmm7[5],xmm4[5],xmm7[6],xmm4[6],xmm7[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3],xmm3[4],xmm4[4],xmm3[5],xmm4[5],xmm3[6],xmm4[6],xmm3[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm3, %xmm6
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm4[4],xmm6[5],xmm4[5],xmm6[6],xmm4[6],xmm6[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1],xmm1[2],xmm4[2],xmm1[3],xmm4[3],xmm1[4],xmm4[4],xmm1[5],xmm4[5],xmm1[6],xmm4[6],xmm1[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm1, %xmm5
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm4[4],xmm5[5],xmm4[5],xmm5[6],xmm4[6],xmm5[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1],xmm1[2],xmm4[2],xmm1[3],xmm4[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1],xmm2[2],xmm4[2],xmm2[3],xmm4[3],xmm2[4],xmm4[4],xmm2[5],xmm4[5],xmm2[6],xmm4[6],xmm2[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm2, %xmm13
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm13 = xmm13[4],xmm4[4],xmm13[5],xmm4[5],xmm13[6],xmm4[6],xmm13[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1],xmm2[2],xmm4[2],xmm2[3],xmm4[3]
; SSE2-NEXT:    paddd %xmm15, %xmm2
; SSE2-NEXT:    paddd %xmm9, %xmm13
; SSE2-NEXT:    paddd %xmm11, %xmm1
; SSE2-NEXT:    paddd %xmm14, %xmm5
; SSE2-NEXT:    paddd %xmm10, %xmm3
; SSE2-NEXT:    paddd %xmm12, %xmm6
; SSE2-NEXT:    paddd %xmm8, %xmm0
; SSE2-NEXT:    paddd -{{[0-9]+}}(%rsp), %xmm7 # 16-byte Folded Reload
; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [1,1,1,1]
; SSE2-NEXT:    paddd %xmm4, %xmm7
; SSE2-NEXT:    paddd %xmm4, %xmm0
; SSE2-NEXT:    paddd %xmm4, %xmm6
; SSE2-NEXT:    paddd %xmm4, %xmm3
; SSE2-NEXT:    paddd %xmm4, %xmm5
; SSE2-NEXT:    paddd %xmm4, %xmm1
; SSE2-NEXT:    paddd %xmm4, %xmm13
; SSE2-NEXT:    paddd %xmm4, %xmm2
; SSE2-NEXT:    psrld $1, %xmm0
; SSE2-NEXT:    psrld $1, %xmm7
; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
; SSE2-NEXT:    pand %xmm4, %xmm7
; SSE2-NEXT:    pand %xmm4, %xmm0
; SSE2-NEXT:    packuswb %xmm7, %xmm0
; SSE2-NEXT:    psrld $1, %xmm3
; SSE2-NEXT:    psrld $1, %xmm6
; SSE2-NEXT:    pand %xmm4, %xmm6
; SSE2-NEXT:    pand %xmm4, %xmm3
; SSE2-NEXT:    packuswb %xmm6, %xmm3
; SSE2-NEXT:    packuswb %xmm3, %xmm0
; SSE2-NEXT:    psrld $1, %xmm1
; SSE2-NEXT:    psrld $1, %xmm5
; SSE2-NEXT:    pand %xmm4, %xmm5
; SSE2-NEXT:    pand %xmm4, %xmm1
; SSE2-NEXT:    packuswb %xmm5, %xmm1
; SSE2-NEXT:    psrld $1, %xmm2
; SSE2-NEXT:    psrld $1, %xmm13
; SSE2-NEXT:    pand %xmm4, %xmm13
; SSE2-NEXT:    pand %xmm4, %xmm2
; SSE2-NEXT:    packuswb %xmm13, %xmm2
; SSE2-NEXT:    packuswb %xmm2, %xmm1
; SSE2-NEXT:    movdqu %xmm1, (%rax)
; SSE2-NEXT:    movdqu %xmm0, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v32i8:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovdqa (%rsi), %ymm0
; AVX2-NEXT:    vpavgb (%rdi), %ymm0, %ymm0
; AVX2-NEXT:    vmovdqu %ymm0, (%rax)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v32i8:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rsi), %ymm0
; AVX512F-NEXT:    vpavgb (%rdi), %ymm0, %ymm0
; AVX512F-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v32i8:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rsi), %ymm0
; AVX512BW-NEXT:    vpavgb (%rdi), %ymm0, %ymm0
; AVX512BW-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <32 x i8>, <32 x i8>* %a
  %2 = load <32 x i8>, <32 x i8>* %b
  %3 = zext <32 x i8> %1 to <32 x i32>
  %4 = zext <32 x i8> %2 to <32 x i32>
  %5 = add nuw nsw <32 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %6 = add nuw nsw <32 x i32> %5, %4
  %7 = lshr <32 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <32 x i32> %7 to <32 x i8>
  store <32 x i8> %8, <32 x i8>* undef, align 4
  ret void
}

define void @avg_v64i8(<64 x i8>* %a, <64 x i8>* %b) {
; SSE2-LABEL: avg_v64i8:
; SSE2:       # BB#0:
; SSE2-NEXT:    subq $152, %rsp
; SSE2-NEXT:  .Lcfi0:
; SSE2-NEXT:    .cfi_def_cfa_offset 160
; SSE2-NEXT:    movdqa (%rdi), %xmm1
; SSE2-NEXT:    movdqa 16(%rdi), %xmm2
; SSE2-NEXT:    movdqa 32(%rdi), %xmm5
; SSE2-NEXT:    movdqa 48(%rdi), %xmm6
; SSE2-NEXT:    pxor %xmm0, %xmm0
; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm1, %xmm3
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm3, {{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm1, {{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1],xmm4[2],xmm0[2],xmm4[3],xmm0[3],xmm4[4],xmm0[4],xmm4[5],xmm0[5],xmm4[6],xmm0[6],xmm4[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm4, %xmm3
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm3, {{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1],xmm4[2],xmm0[2],xmm4[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm4, {{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    movdqa %xmm2, %xmm1
; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm1, %xmm2
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm2, {{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm1, {{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm3, %xmm2
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm2, {{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm3, {{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm5[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1],xmm5[2],xmm0[2],xmm5[3],xmm0[3],xmm5[4],xmm0[4],xmm5[5],xmm0[5],xmm5[6],xmm0[6],xmm5[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm5, %xmm2
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm2, (%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1],xmm5[2],xmm0[2],xmm5[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm5, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm3, %xmm2
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm2, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm3, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm6[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1],xmm6[2],xmm0[2],xmm6[3],xmm0[3],xmm6[4],xmm0[4],xmm6[5],xmm0[5],xmm6[6],xmm0[6],xmm6[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm6, %xmm1
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm1 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1],xmm6[2],xmm0[2],xmm6[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm6, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm9 = xmm9[0],xmm0[0],xmm9[1],xmm0[1],xmm9[2],xmm0[2],xmm9[3],xmm0[3],xmm9[4],xmm0[4],xmm9[5],xmm0[5],xmm9[6],xmm0[6],xmm9[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm9, %xmm1
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm1 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm9 = xmm9[0],xmm0[0],xmm9[1],xmm0[1],xmm9[2],xmm0[2],xmm9[3],xmm0[3]
; SSE2-NEXT:    movdqa (%rsi), %xmm15
; SSE2-NEXT:    pshufd {{.*#+}} xmm7 = xmm15[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm0[0],xmm15[1],xmm0[1],xmm15[2],xmm0[2],xmm15[3],xmm0[3],xmm15[4],xmm0[4],xmm15[5],xmm0[5],xmm15[6],xmm0[6],xmm15[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm15, %xmm10
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm10 = xmm10[4],xmm0[4],xmm10[5],xmm0[5],xmm10[6],xmm0[6],xmm10[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm15 = xmm15[0],xmm0[0],xmm15[1],xmm0[1],xmm15[2],xmm0[2],xmm15[3],xmm0[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm0[0],xmm7[1],xmm0[1],xmm7[2],xmm0[2],xmm7[3],xmm0[3],xmm7[4],xmm0[4],xmm7[5],xmm0[5],xmm7[6],xmm0[6],xmm7[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm7, %xmm14
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm14 = xmm14[4],xmm0[4],xmm14[5],xmm0[5],xmm14[6],xmm0[6],xmm14[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm0[0],xmm7[1],xmm0[1],xmm7[2],xmm0[2],xmm7[3],xmm0[3]
; SSE2-NEXT:    movdqa 16(%rsi), %xmm1
; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm1[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm1, %xmm13
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm13 = xmm13[4],xmm0[4],xmm13[5],xmm0[5],xmm13[6],xmm0[6],xmm13[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1],xmm6[2],xmm0[2],xmm6[3],xmm0[3],xmm6[4],xmm0[4],xmm6[5],xmm0[5],xmm6[6],xmm0[6],xmm6[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm6, %xmm12
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm12 = xmm12[4],xmm0[4],xmm12[5],xmm0[5],xmm12[6],xmm0[6],xmm12[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1],xmm6[2],xmm0[2],xmm6[3],xmm0[3]
; SSE2-NEXT:    movdqa 32(%rsi), %xmm2
; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm2[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm2, %xmm11
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm11 = xmm11[4],xmm0[4],xmm11[5],xmm0[5],xmm11[6],xmm0[6],xmm11[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1],xmm5[2],xmm0[2],xmm5[3],xmm0[3],xmm5[4],xmm0[4],xmm5[5],xmm0[5],xmm5[6],xmm0[6],xmm5[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm5, %xmm3
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm3, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1],xmm5[2],xmm0[2],xmm5[3],xmm0[3]
; SSE2-NEXT:    movdqa 48(%rsi), %xmm3
; SSE2-NEXT:    pshufd {{.*#+}} xmm8 = xmm3[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm3, %xmm4
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm0[4],xmm4[5],xmm0[5],xmm4[6],xmm0[6],xmm4[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm4, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm8 = xmm8[0],xmm0[0],xmm8[1],xmm0[1],xmm8[2],xmm0[2],xmm8[3],xmm0[3],xmm8[4],xmm0[4],xmm8[5],xmm0[5],xmm8[6],xmm0[6],xmm8[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm8, %xmm4
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm0[4],xmm4[5],xmm0[5],xmm4[6],xmm0[6],xmm4[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm0[0],xmm8[1],xmm0[1],xmm8[2],xmm0[2],xmm8[3],xmm0[3]
; SSE2-NEXT:    paddd %xmm9, %xmm8
; SSE2-NEXT:    paddd -{{[0-9]+}}(%rsp), %xmm4 # 16-byte Folded Reload
; SSE2-NEXT:    movdqa %xmm4, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    paddd -{{[0-9]+}}(%rsp), %xmm3 # 16-byte Folded Reload
; SSE2-NEXT:    movdqa -{{[0-9]+}}(%rsp), %xmm9 # 16-byte Reload
; SSE2-NEXT:    paddd -{{[0-9]+}}(%rsp), %xmm9 # 16-byte Folded Reload
; SSE2-NEXT:    paddd -{{[0-9]+}}(%rsp), %xmm5 # 16-byte Folded Reload
; SSE2-NEXT:    movdqa -{{[0-9]+}}(%rsp), %xmm4 # 16-byte Reload
; SSE2-NEXT:    paddd -{{[0-9]+}}(%rsp), %xmm4 # 16-byte Folded Reload
; SSE2-NEXT:    paddd -{{[0-9]+}}(%rsp), %xmm2 # 16-byte Folded Reload
; SSE2-NEXT:    paddd (%rsp), %xmm11 # 16-byte Folded Reload
; SSE2-NEXT:    paddd {{[0-9]+}}(%rsp), %xmm6 # 16-byte Folded Reload
; SSE2-NEXT:    paddd {{[0-9]+}}(%rsp), %xmm12 # 16-byte Folded Reload
; SSE2-NEXT:    paddd {{[0-9]+}}(%rsp), %xmm1 # 16-byte Folded Reload
; SSE2-NEXT:    paddd {{[0-9]+}}(%rsp), %xmm13 # 16-byte Folded Reload
; SSE2-NEXT:    paddd {{[0-9]+}}(%rsp), %xmm7 # 16-byte Folded Reload
; SSE2-NEXT:    paddd {{[0-9]+}}(%rsp), %xmm14 # 16-byte Folded Reload
; SSE2-NEXT:    paddd {{[0-9]+}}(%rsp), %xmm15 # 16-byte Folded Reload
; SSE2-NEXT:    paddd {{[0-9]+}}(%rsp), %xmm10 # 16-byte Folded Reload
; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [1,1,1,1]
; SSE2-NEXT:    paddd %xmm0, %xmm10
; SSE2-NEXT:    paddd %xmm0, %xmm15
; SSE2-NEXT:    paddd %xmm0, %xmm14
; SSE2-NEXT:    paddd %xmm0, %xmm7
; SSE2-NEXT:    paddd %xmm0, %xmm13
; SSE2-NEXT:    paddd %xmm0, %xmm1
; SSE2-NEXT:    paddd %xmm0, %xmm12
; SSE2-NEXT:    paddd %xmm0, %xmm6
; SSE2-NEXT:    paddd %xmm0, %xmm11
; SSE2-NEXT:    paddd %xmm0, %xmm2
; SSE2-NEXT:    paddd %xmm0, %xmm4
; SSE2-NEXT:    movdqa %xmm4, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    paddd %xmm0, %xmm5
; SSE2-NEXT:    paddd %xmm0, %xmm9
; SSE2-NEXT:    paddd %xmm0, %xmm3
; SSE2-NEXT:    movdqa -{{[0-9]+}}(%rsp), %xmm4 # 16-byte Reload
; SSE2-NEXT:    paddd %xmm0, %xmm4
; SSE2-NEXT:    paddd %xmm0, %xmm8
; SSE2-NEXT:    psrld $1, %xmm15
; SSE2-NEXT:    psrld $1, %xmm10
; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
; SSE2-NEXT:    pand %xmm0, %xmm10
; SSE2-NEXT:    pand %xmm0, %xmm15
; SSE2-NEXT:    packuswb %xmm10, %xmm15
; SSE2-NEXT:    psrld $1, %xmm7
; SSE2-NEXT:    psrld $1, %xmm14
; SSE2-NEXT:    pand %xmm0, %xmm14
; SSE2-NEXT:    pand %xmm0, %xmm7
; SSE2-NEXT:    packuswb %xmm14, %xmm7
; SSE2-NEXT:    packuswb %xmm7, %xmm15
; SSE2-NEXT:    psrld $1, %xmm1
; SSE2-NEXT:    psrld $1, %xmm13
; SSE2-NEXT:    pand %xmm0, %xmm13
; SSE2-NEXT:    pand %xmm0, %xmm1
; SSE2-NEXT:    packuswb %xmm13, %xmm1
; SSE2-NEXT:    psrld $1, %xmm6
; SSE2-NEXT:    psrld $1, %xmm12
; SSE2-NEXT:    pand %xmm0, %xmm12
; SSE2-NEXT:    pand %xmm0, %xmm6
; SSE2-NEXT:    packuswb %xmm12, %xmm6
; SSE2-NEXT:    packuswb %xmm6, %xmm1
; SSE2-NEXT:    psrld $1, %xmm2
; SSE2-NEXT:    psrld $1, %xmm11
; SSE2-NEXT:    pand %xmm0, %xmm11
; SSE2-NEXT:    pand %xmm0, %xmm2
; SSE2-NEXT:    packuswb %xmm11, %xmm2
; SSE2-NEXT:    psrld $1, %xmm5
; SSE2-NEXT:    movdqa -{{[0-9]+}}(%rsp), %xmm6 # 16-byte Reload
; SSE2-NEXT:    psrld $1, %xmm6
; SSE2-NEXT:    pand %xmm0, %xmm6
; SSE2-NEXT:    pand %xmm0, %xmm5
; SSE2-NEXT:    packuswb %xmm6, %xmm5
; SSE2-NEXT:    packuswb %xmm5, %xmm2
; SSE2-NEXT:    psrld $1, %xmm3
; SSE2-NEXT:    movdqa %xmm9, %xmm5
; SSE2-NEXT:    psrld $1, %xmm5
; SSE2-NEXT:    pand %xmm0, %xmm5
; SSE2-NEXT:    pand %xmm0, %xmm3
; SSE2-NEXT:    packuswb %xmm5, %xmm3
; SSE2-NEXT:    psrld $1, %xmm8
; SSE2-NEXT:    movdqa %xmm4, %xmm5
; SSE2-NEXT:    psrld $1, %xmm5
; SSE2-NEXT:    pand %xmm0, %xmm5
; SSE2-NEXT:    pand %xmm0, %xmm8
; SSE2-NEXT:    packuswb %xmm5, %xmm8
; SSE2-NEXT:    packuswb %xmm8, %xmm3
; SSE2-NEXT:    movdqu %xmm3, (%rax)
; SSE2-NEXT:    movdqu %xmm2, (%rax)
; SSE2-NEXT:    movdqu %xmm1, (%rax)
; SSE2-NEXT:    movdqu %xmm15, (%rax)
; SSE2-NEXT:    addq $152, %rsp
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v64i8:
; AVX2:       # BB#0:
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm5 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm6 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm8 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm9 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm10 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm11 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm12 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm13 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm14 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm15 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpaddd %ymm15, %ymm7, %ymm7
; AVX2-NEXT:    vpaddd %ymm14, %ymm6, %ymm6
; AVX2-NEXT:    vpaddd %ymm13, %ymm5, %ymm5
; AVX2-NEXT:    vpaddd %ymm12, %ymm4, %ymm4
; AVX2-NEXT:    vpaddd %ymm11, %ymm3, %ymm3
; AVX2-NEXT:    vpaddd %ymm10, %ymm2, %ymm2
; AVX2-NEXT:    vpaddd %ymm9, %ymm1, %ymm1
; AVX2-NEXT:    vpaddd %ymm8, %ymm0, %ymm0
; AVX2-NEXT:    vpbroadcastd {{.*}}(%rip), %ymm8
; AVX2-NEXT:    vpaddd %ymm8, %ymm0, %ymm9
; AVX2-NEXT:    vpaddd %ymm8, %ymm1, %ymm10
; AVX2-NEXT:    vpaddd %ymm8, %ymm2, %ymm2
; AVX2-NEXT:    vpaddd %ymm8, %ymm3, %ymm3
; AVX2-NEXT:    vpaddd %ymm8, %ymm4, %ymm4
; AVX2-NEXT:    vpaddd %ymm8, %ymm5, %ymm5
; AVX2-NEXT:    vpaddd %ymm8, %ymm6, %ymm1
; AVX2-NEXT:    vpaddd %ymm8, %ymm7, %ymm0
; AVX2-NEXT:    vpsrld $1, %ymm0, %ymm11
; AVX2-NEXT:    vpsrld $1, %ymm1, %ymm12
; AVX2-NEXT:    vpsrld $1, %ymm5, %ymm5
; AVX2-NEXT:    vpsrld $1, %ymm4, %ymm4
; AVX2-NEXT:    vpsrld $1, %ymm3, %ymm6
; AVX2-NEXT:    vpsrld $1, %ymm2, %ymm7
; AVX2-NEXT:    vpsrld $1, %ymm10, %ymm8
; AVX2-NEXT:    vpsrld $1, %ymm9, %ymm3
; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,1,4,5,8,9,12,13,128,128,128,128,128,128,128,128,0,1,4,5,8,9,12,13,128,128,128,128,128,128,128,128]
; AVX2-NEXT:    vpshufb %ymm2, %ymm3, %ymm3
; AVX2-NEXT:    vpermq {{.*#+}} ymm9 = ymm3[0,2,2,3]
; AVX2-NEXT:    vmovdqa {{.*#+}} xmm3 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
; AVX2-NEXT:    vpshufb %xmm3, %xmm9, %xmm0
; AVX2-NEXT:    vpshufb %ymm2, %ymm8, %ymm8
; AVX2-NEXT:    vpermq {{.*#+}} ymm8 = ymm8[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm8, %xmm1
; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; AVX2-NEXT:    vpshufb %ymm2, %ymm7, %ymm1
; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
; AVX2-NEXT:    vpshufb %ymm2, %ymm6, %ymm6
; AVX2-NEXT:    vpermq {{.*#+}} ymm6 = ymm6[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm6, %xmm6
; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm6[0],xmm1[0]
; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
; AVX2-NEXT:    vpshufb %ymm2, %ymm4, %ymm1
; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
; AVX2-NEXT:    vpshufb %ymm2, %ymm5, %ymm4
; AVX2-NEXT:    vpermq {{.*#+}} ymm4 = ymm4[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm4, %xmm4
; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm4[0],xmm1[0]
; AVX2-NEXT:    vpshufb %ymm2, %ymm12, %ymm4
; AVX2-NEXT:    vpermq {{.*#+}} ymm4 = ymm4[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm4, %xmm4
; AVX2-NEXT:    vpshufb %ymm2, %ymm11, %ymm2
; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm2, %xmm2
; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm4[0]
; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm2, %ymm1
; AVX2-NEXT:    vmovdqu %ymm1, (%rax)
; AVX2-NEXT:    vmovdqu %ymm0, (%rax)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v64i8:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm5 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm6 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpaddd %zmm7, %zmm3, %zmm3
; AVX512F-NEXT:    vpaddd %zmm6, %zmm2, %zmm2
; AVX512F-NEXT:    vpaddd %zmm5, %zmm1, %zmm1
; AVX512F-NEXT:    vpaddd %zmm4, %zmm0, %zmm0
; AVX512F-NEXT:    vpbroadcastd {{.*}}(%rip), %zmm4
; AVX512F-NEXT:    vpaddd %zmm4, %zmm0, %zmm0
; AVX512F-NEXT:    vpaddd %zmm4, %zmm1, %zmm1
; AVX512F-NEXT:    vpaddd %zmm4, %zmm2, %zmm2
; AVX512F-NEXT:    vpaddd %zmm4, %zmm3, %zmm3
; AVX512F-NEXT:    vpsrld $1, %zmm3, %zmm3
; AVX512F-NEXT:    vpsrld $1, %zmm2, %zmm2
; AVX512F-NEXT:    vpsrld $1, %zmm1, %zmm1
; AVX512F-NEXT:    vpsrld $1, %zmm0, %zmm0
; AVX512F-NEXT:    vpmovdb %zmm0, %xmm0
; AVX512F-NEXT:    vpmovdb %zmm1, %xmm1
; AVX512F-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX512F-NEXT:    vpmovdb %zmm2, %xmm1
; AVX512F-NEXT:    vpmovdb %zmm3, %xmm2
; AVX512F-NEXT:    vinserti128 $1, %xmm2, %ymm1, %ymm1
; AVX512F-NEXT:    vmovdqu %ymm1, (%rax)
; AVX512F-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v64i8:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqu8 (%rsi), %zmm0
; AVX512BW-NEXT:    vpavgb (%rdi), %zmm0, %zmm0
; AVX512BW-NEXT:    vmovdqu8 %zmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <64 x i8>, <64 x i8>* %a
  %2 = load <64 x i8>, <64 x i8>* %b
  %3 = zext <64 x i8> %1 to <64 x i32>
  %4 = zext <64 x i8> %2 to <64 x i32>
  %5 = add nuw nsw <64 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %6 = add nuw nsw <64 x i32> %5, %4
  %7 = lshr <64 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <64 x i32> %7 to <64 x i8>
  store <64 x i8> %8, <64 x i8>* undef, align 4
  ret void
}

define void @avg_v4i16(<4 x i16>* %a, <4 x i16>* %b) {
; SSE2-LABEL: avg_v4i16:
; SSE2:       # BB#0:
; SSE2-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
; SSE2-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
; SSE2-NEXT:    pavgw %xmm0, %xmm1
; SSE2-NEXT:    movq %xmm1, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v4i16:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
; AVX2-NEXT:    vpavgw %xmm0, %xmm1, %xmm0
; AVX2-NEXT:    vmovq %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v4i16:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512F-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
; AVX512F-NEXT:    vpavgw %xmm0, %xmm1, %xmm0
; AVX512F-NEXT:    vmovq %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v4i16:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512BW-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
; AVX512BW-NEXT:    vpavgw %xmm0, %xmm1, %xmm0
; AVX512BW-NEXT:    vmovq %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <4 x i16>, <4 x i16>* %a
  %2 = load <4 x i16>, <4 x i16>* %b
  %3 = zext <4 x i16> %1 to <4 x i32>
  %4 = zext <4 x i16> %2 to <4 x i32>
  %5 = add nuw nsw <4 x i32> %3, <i32 1, i32 1, i32 1, i32 1>
  %6 = add nuw nsw <4 x i32> %5, %4
  %7 = lshr <4 x i32> %6, <i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <4 x i32> %7 to <4 x i16>
  store <4 x i16> %8, <4 x i16>* undef, align 4
  ret void
}

define void @avg_v8i16(<8 x i16>* %a, <8 x i16>* %b) {
; SSE2-LABEL: avg_v8i16:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rsi), %xmm0
; SSE2-NEXT:    pavgw (%rdi), %xmm0
; SSE2-NEXT:    movdqu %xmm0, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v8i16:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovdqa (%rsi), %xmm0
; AVX2-NEXT:    vpavgw (%rdi), %xmm0, %xmm0
; AVX2-NEXT:    vmovdqu %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v8i16:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rsi), %xmm0
; AVX512F-NEXT:    vpavgw (%rdi), %xmm0, %xmm0
; AVX512F-NEXT:    vmovdqu %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v8i16:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rsi), %xmm0
; AVX512BW-NEXT:    vpavgw (%rdi), %xmm0, %xmm0
; AVX512BW-NEXT:    vmovdqu %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <8 x i16>, <8 x i16>* %a
  %2 = load <8 x i16>, <8 x i16>* %b
  %3 = zext <8 x i16> %1 to <8 x i32>
  %4 = zext <8 x i16> %2 to <8 x i32>
  %5 = add nuw nsw <8 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %6 = add nuw nsw <8 x i32> %5, %4
  %7 = lshr <8 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <8 x i32> %7 to <8 x i16>
  store <8 x i16> %8, <8 x i16>* undef, align 4
  ret void
}

define void @avg_v16i16(<16 x i16>* %a, <16 x i16>* %b) {
; SSE2-LABEL: avg_v16i16:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm4
; SSE2-NEXT:    movdqa 16(%rdi), %xmm5
; SSE2-NEXT:    movdqa (%rsi), %xmm0
; SSE2-NEXT:    movdqa 16(%rsi), %xmm1
; SSE2-NEXT:    pxor %xmm6, %xmm6
; SSE2-NEXT:    movdqa %xmm4, %xmm8
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm8 = xmm8[4],xmm6[4],xmm8[5],xmm6[5],xmm8[6],xmm6[6],xmm8[7],xmm6[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm6[0],xmm4[1],xmm6[1],xmm4[2],xmm6[2],xmm4[3],xmm6[3]
; SSE2-NEXT:    movdqa %xmm5, %xmm7
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm7 = xmm7[4],xmm6[4],xmm7[5],xmm6[5],xmm7[6],xmm6[6],xmm7[7],xmm6[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm6[0],xmm5[1],xmm6[1],xmm5[2],xmm6[2],xmm5[3],xmm6[3]
; SSE2-NEXT:    movdqa %xmm0, %xmm3
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm6[4],xmm3[5],xmm6[5],xmm3[6],xmm6[6],xmm3[7],xmm6[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm6[0],xmm0[1],xmm6[1],xmm0[2],xmm6[2],xmm0[3],xmm6[3]
; SSE2-NEXT:    movdqa %xmm1, %xmm2
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm6[4],xmm2[5],xmm6[5],xmm2[6],xmm6[6],xmm2[7],xmm6[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm6[0],xmm1[1],xmm6[1],xmm1[2],xmm6[2],xmm1[3],xmm6[3]
; SSE2-NEXT:    paddd %xmm5, %xmm1
; SSE2-NEXT:    paddd %xmm7, %xmm2
; SSE2-NEXT:    paddd %xmm4, %xmm0
; SSE2-NEXT:    paddd %xmm8, %xmm3
; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [1,1,1,1]
; SSE2-NEXT:    paddd %xmm4, %xmm3
; SSE2-NEXT:    paddd %xmm4, %xmm0
; SSE2-NEXT:    paddd %xmm4, %xmm2
; SSE2-NEXT:    paddd %xmm4, %xmm1
; SSE2-NEXT:    psrld $1, %xmm1
; SSE2-NEXT:    psrld $1, %xmm2
; SSE2-NEXT:    psrld $1, %xmm0
; SSE2-NEXT:    psrld $1, %xmm3
; SSE2-NEXT:    pslld $16, %xmm3
; SSE2-NEXT:    psrad $16, %xmm3
; SSE2-NEXT:    pslld $16, %xmm0
; SSE2-NEXT:    psrad $16, %xmm0
; SSE2-NEXT:    packssdw %xmm3, %xmm0
; SSE2-NEXT:    pslld $16, %xmm2
; SSE2-NEXT:    psrad $16, %xmm2
; SSE2-NEXT:    pslld $16, %xmm1
; SSE2-NEXT:    psrad $16, %xmm1
; SSE2-NEXT:    packssdw %xmm2, %xmm1
; SSE2-NEXT:    movdqu %xmm1, (%rax)
; SSE2-NEXT:    movdqu %xmm0, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v16i16:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovdqa (%rsi), %ymm0
; AVX2-NEXT:    vpavgw (%rdi), %ymm0, %ymm0
; AVX2-NEXT:    vmovdqu %ymm0, (%rax)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v16i16:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rsi), %ymm0
; AVX512F-NEXT:    vpavgw (%rdi), %ymm0, %ymm0
; AVX512F-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v16i16:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rsi), %ymm0
; AVX512BW-NEXT:    vpavgw (%rdi), %ymm0, %ymm0
; AVX512BW-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <16 x i16>, <16 x i16>* %a
  %2 = load <16 x i16>, <16 x i16>* %b
  %3 = zext <16 x i16> %1 to <16 x i32>
  %4 = zext <16 x i16> %2 to <16 x i32>
  %5 = add nuw nsw <16 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %6 = add nuw nsw <16 x i32> %5, %4
  %7 = lshr <16 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <16 x i32> %7 to <16 x i16>
  store <16 x i16> %8, <16 x i16>* undef, align 4
  ret void
}

define void @avg_v32i16(<32 x i16>* %a, <32 x i16>* %b) {
; SSE2-LABEL: avg_v32i16:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm10
; SSE2-NEXT:    movdqa 16(%rdi), %xmm9
; SSE2-NEXT:    movdqa 32(%rdi), %xmm11
; SSE2-NEXT:    movdqa 48(%rdi), %xmm8
; SSE2-NEXT:    movdqa (%rsi), %xmm14
; SSE2-NEXT:    movdqa 16(%rsi), %xmm1
; SSE2-NEXT:    movdqa 32(%rsi), %xmm2
; SSE2-NEXT:    movdqa 48(%rsi), %xmm3
; SSE2-NEXT:    pxor %xmm0, %xmm0
; SSE2-NEXT:    movdqa %xmm10, %xmm4
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm0[4],xmm4[5],xmm0[5],xmm4[6],xmm0[6],xmm4[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm4, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm10 = xmm10[0],xmm0[0],xmm10[1],xmm0[1],xmm10[2],xmm0[2],xmm10[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm9, %xmm12
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm12 = xmm12[4],xmm0[4],xmm12[5],xmm0[5],xmm12[6],xmm0[6],xmm12[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm9 = xmm9[0],xmm0[0],xmm9[1],xmm0[1],xmm9[2],xmm0[2],xmm9[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm11, %xmm15
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm15 = xmm15[4],xmm0[4],xmm15[5],xmm0[5],xmm15[6],xmm0[6],xmm15[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm11 = xmm11[0],xmm0[0],xmm11[1],xmm0[1],xmm11[2],xmm0[2],xmm11[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm8, %xmm13
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm13 = xmm13[4],xmm0[4],xmm13[5],xmm0[5],xmm13[6],xmm0[6],xmm13[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm0[0],xmm8[1],xmm0[1],xmm8[2],xmm0[2],xmm8[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm14, %xmm7
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm7 = xmm7[4],xmm0[4],xmm7[5],xmm0[5],xmm7[6],xmm0[6],xmm7[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm14 = xmm14[0],xmm0[0],xmm14[1],xmm0[1],xmm14[2],xmm0[2],xmm14[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm1, %xmm6
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm0[4],xmm6[5],xmm0[5],xmm6[6],xmm0[6],xmm6[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm2, %xmm5
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm0[4],xmm5[5],xmm0[5],xmm5[6],xmm0[6],xmm5[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm3, %xmm4
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm0[4],xmm4[5],xmm0[5],xmm4[6],xmm0[6],xmm4[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
; SSE2-NEXT:    paddd %xmm8, %xmm3
; SSE2-NEXT:    paddd %xmm13, %xmm4
; SSE2-NEXT:    paddd %xmm11, %xmm2
; SSE2-NEXT:    paddd %xmm15, %xmm5
; SSE2-NEXT:    paddd %xmm9, %xmm1
; SSE2-NEXT:    paddd %xmm12, %xmm6
; SSE2-NEXT:    paddd %xmm10, %xmm14
; SSE2-NEXT:    paddd -{{[0-9]+}}(%rsp), %xmm7 # 16-byte Folded Reload
; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [1,1,1,1]
; SSE2-NEXT:    paddd %xmm0, %xmm7
; SSE2-NEXT:    paddd %xmm0, %xmm14
; SSE2-NEXT:    paddd %xmm0, %xmm6
; SSE2-NEXT:    paddd %xmm0, %xmm1
; SSE2-NEXT:    paddd %xmm0, %xmm5
; SSE2-NEXT:    paddd %xmm0, %xmm2
; SSE2-NEXT:    paddd %xmm0, %xmm4
; SSE2-NEXT:    paddd %xmm0, %xmm3
; SSE2-NEXT:    psrld $1, %xmm14
; SSE2-NEXT:    psrld $1, %xmm7
; SSE2-NEXT:    pslld $16, %xmm7
; SSE2-NEXT:    psrad $16, %xmm7
; SSE2-NEXT:    pslld $16, %xmm14
; SSE2-NEXT:    psrad $16, %xmm14
; SSE2-NEXT:    packssdw %xmm7, %xmm14
; SSE2-NEXT:    psrld $1, %xmm1
; SSE2-NEXT:    psrld $1, %xmm6
; SSE2-NEXT:    pslld $16, %xmm6
; SSE2-NEXT:    psrad $16, %xmm6
; SSE2-NEXT:    pslld $16, %xmm1
; SSE2-NEXT:    psrad $16, %xmm1
; SSE2-NEXT:    packssdw %xmm6, %xmm1
; SSE2-NEXT:    psrld $1, %xmm2
; SSE2-NEXT:    psrld $1, %xmm5
; SSE2-NEXT:    pslld $16, %xmm5
; SSE2-NEXT:    psrad $16, %xmm5
; SSE2-NEXT:    pslld $16, %xmm2
; SSE2-NEXT:    psrad $16, %xmm2
; SSE2-NEXT:    packssdw %xmm5, %xmm2
; SSE2-NEXT:    psrld $1, %xmm3
; SSE2-NEXT:    psrld $1, %xmm4
; SSE2-NEXT:    pslld $16, %xmm4
; SSE2-NEXT:    psrad $16, %xmm4
; SSE2-NEXT:    pslld $16, %xmm3
; SSE2-NEXT:    psrad $16, %xmm3
; SSE2-NEXT:    packssdw %xmm4, %xmm3
; SSE2-NEXT:    movdqu %xmm3, (%rax)
; SSE2-NEXT:    movdqu %xmm2, (%rax)
; SSE2-NEXT:    movdqu %xmm1, (%rax)
; SSE2-NEXT:    movdqu %xmm14, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v32i16:
; AVX2:       # BB#0:
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm2 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm4 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm5 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm6 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpaddd %ymm7, %ymm3, %ymm3
; AVX2-NEXT:    vpaddd %ymm6, %ymm2, %ymm2
; AVX2-NEXT:    vpaddd %ymm5, %ymm1, %ymm1
; AVX2-NEXT:    vpaddd %ymm4, %ymm0, %ymm0
; AVX2-NEXT:    vpbroadcastd {{.*}}(%rip), %ymm4
; AVX2-NEXT:    vpaddd %ymm4, %ymm0, %ymm0
; AVX2-NEXT:    vpaddd %ymm4, %ymm1, %ymm1
; AVX2-NEXT:    vpaddd %ymm4, %ymm2, %ymm2
; AVX2-NEXT:    vpaddd %ymm4, %ymm3, %ymm3
; AVX2-NEXT:    vpsrld $1, %ymm3, %ymm3
; AVX2-NEXT:    vpsrld $1, %ymm2, %ymm2
; AVX2-NEXT:    vpsrld $1, %ymm1, %ymm1
; AVX2-NEXT:    vpsrld $1, %ymm0, %ymm0
; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = [0,1,4,5,8,9,12,13,128,128,128,128,128,128,128,128,0,1,4,5,8,9,12,13,128,128,128,128,128,128,128,128]
; AVX2-NEXT:    vpshufb %ymm4, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
; AVX2-NEXT:    vpshufb %ymm4, %ymm1, %ymm1
; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT:    vpshufb %ymm4, %ymm2, %ymm1
; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
; AVX2-NEXT:    vpshufb %ymm4, %ymm3, %ymm2
; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
; AVX2-NEXT:    vinserti128 $1, %xmm2, %ymm1, %ymm1
; AVX2-NEXT:    vmovdqu %ymm1, (%rax)
; AVX2-NEXT:    vmovdqu %ymm0, (%rax)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v32i16:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vpmovzxwd {{.*#+}} zmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero
; AVX512F-NEXT:    vpmovzxwd {{.*#+}} zmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero
; AVX512F-NEXT:    vpmovzxwd {{.*#+}} zmm2 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero
; AVX512F-NEXT:    vpmovzxwd {{.*#+}} zmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero
; AVX512F-NEXT:    vpaddd %zmm3, %zmm1, %zmm1
; AVX512F-NEXT:    vpaddd %zmm2, %zmm0, %zmm0
; AVX512F-NEXT:    vpbroadcastd {{.*}}(%rip), %zmm2
; AVX512F-NEXT:    vpaddd %zmm2, %zmm0, %zmm0
; AVX512F-NEXT:    vpaddd %zmm2, %zmm1, %zmm1
; AVX512F-NEXT:    vpsrld $1, %zmm1, %zmm1
; AVX512F-NEXT:    vpsrld $1, %zmm0, %zmm0
; AVX512F-NEXT:    vpmovdw %zmm0, (%rax)
; AVX512F-NEXT:    vpmovdw %zmm1, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v32i16:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqu16 (%rsi), %zmm0
; AVX512BW-NEXT:    vpavgw (%rdi), %zmm0, %zmm0
; AVX512BW-NEXT:    vmovdqu16 %zmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <32 x i16>, <32 x i16>* %a
  %2 = load <32 x i16>, <32 x i16>* %b
  %3 = zext <32 x i16> %1 to <32 x i32>
  %4 = zext <32 x i16> %2 to <32 x i32>
  %5 = add nuw nsw <32 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %6 = add nuw nsw <32 x i32> %5, %4
  %7 = lshr <32 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <32 x i32> %7 to <32 x i16>
  store <32 x i16> %8, <32 x i16>* undef, align 4
  ret void
}

define void @avg_v4i8_2(<4 x i8>* %a, <4 x i8>* %b) {
; SSE2-LABEL: avg_v4i8_2:
; SSE2:       # BB#0:
; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; SSE2-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
; SSE2-NEXT:    pavgb %xmm0, %xmm1
; SSE2-NEXT:    movd %xmm1, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v4i8_2:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX2-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
; AVX2-NEXT:    vpavgb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vmovd %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v4i8_2:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX512F-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
; AVX512F-NEXT:    vpavgb %xmm1, %xmm0, %xmm0
; AVX512F-NEXT:    vmovd %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v4i8_2:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX512BW-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
; AVX512BW-NEXT:    vpavgb %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT:    vmovd %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <4 x i8>, <4 x i8>* %a
  %2 = load <4 x i8>, <4 x i8>* %b
  %3 = zext <4 x i8> %1 to <4 x i32>
  %4 = zext <4 x i8> %2 to <4 x i32>
  %5 = add nuw nsw <4 x i32> %3, %4
  %6 = add nuw nsw <4 x i32> %5, <i32 1, i32 1, i32 1, i32 1>
  %7 = lshr <4 x i32> %6, <i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <4 x i32> %7 to <4 x i8>
  store <4 x i8> %8, <4 x i8>* undef, align 4
  ret void
}

define void @avg_v8i8_2(<8 x i8>* %a, <8 x i8>* %b) {
; SSE2-LABEL: avg_v8i8_2:
; SSE2:       # BB#0:
; SSE2-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
; SSE2-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
; SSE2-NEXT:    pavgb %xmm0, %xmm1
; SSE2-NEXT:    movq %xmm1, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v8i8_2:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
; AVX2-NEXT:    vpavgb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vmovq %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v8i8_2:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512F-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
; AVX512F-NEXT:    vpavgb %xmm1, %xmm0, %xmm0
; AVX512F-NEXT:    vmovq %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v8i8_2:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512BW-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
; AVX512BW-NEXT:    vpavgb %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT:    vmovq %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <8 x i8>, <8 x i8>* %a
  %2 = load <8 x i8>, <8 x i8>* %b
  %3 = zext <8 x i8> %1 to <8 x i32>
  %4 = zext <8 x i8> %2 to <8 x i32>
  %5 = add nuw nsw <8 x i32> %3, %4
  %6 = add nuw nsw <8 x i32> %5, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %7 = lshr <8 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <8 x i32> %7 to <8 x i8>
  store <8 x i8> %8, <8 x i8>* undef, align 4
  ret void
}

define void @avg_v16i8_2(<16 x i8>* %a, <16 x i8>* %b) {
; SSE2-LABEL: avg_v16i8_2:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pavgb (%rsi), %xmm0
; SSE2-NEXT:    movdqu %xmm0, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v16i8_2:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovdqa (%rdi), %xmm0
; AVX2-NEXT:    vpavgb (%rsi), %xmm0, %xmm0
; AVX2-NEXT:    vmovdqu %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v16i8_2:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpavgb (%rsi), %xmm0, %xmm0
; AVX512F-NEXT:    vmovdqu %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v16i8_2:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpavgb (%rsi), %xmm0, %xmm0
; AVX512BW-NEXT:    vmovdqu %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <16 x i8>, <16 x i8>* %a
  %2 = load <16 x i8>, <16 x i8>* %b
  %3 = zext <16 x i8> %1 to <16 x i32>
  %4 = zext <16 x i8> %2 to <16 x i32>
  %5 = add nuw nsw <16 x i32> %3, %4
  %6 = add nuw nsw <16 x i32> %5, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %7 = lshr <16 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <16 x i32> %7 to <16 x i8>
  store <16 x i8> %8, <16 x i8>* undef, align 4
  ret void
}

define void @avg_v32i8_2(<32 x i8>* %a, <32 x i8>* %b) {
; SSE2-LABEL: avg_v32i8_2:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm8
; SSE2-NEXT:    movdqa 16(%rdi), %xmm11
; SSE2-NEXT:    movdqa (%rsi), %xmm0
; SSE2-NEXT:    movdqa 16(%rsi), %xmm1
; SSE2-NEXT:    pxor %xmm4, %xmm4
; SSE2-NEXT:    pshufd {{.*#+}} xmm10 = xmm8[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm8 = xmm8[0],xmm4[0],xmm8[1],xmm4[1],xmm8[2],xmm4[2],xmm8[3],xmm4[3],xmm8[4],xmm4[4],xmm8[5],xmm4[5],xmm8[6],xmm4[6],xmm8[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm8, %xmm2
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm4[4],xmm2[5],xmm4[5],xmm2[6],xmm4[6],xmm2[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm2, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm4[0],xmm8[1],xmm4[1],xmm8[2],xmm4[2],xmm8[3],xmm4[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm10 = xmm10[0],xmm4[0],xmm10[1],xmm4[1],xmm10[2],xmm4[2],xmm10[3],xmm4[3],xmm10[4],xmm4[4],xmm10[5],xmm4[5],xmm10[6],xmm4[6],xmm10[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm10, %xmm12
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm12 = xmm12[4],xmm4[4],xmm12[5],xmm4[5],xmm12[6],xmm4[6],xmm12[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm10 = xmm10[0],xmm4[0],xmm10[1],xmm4[1],xmm10[2],xmm4[2],xmm10[3],xmm4[3]
; SSE2-NEXT:    pshufd {{.*#+}} xmm15 = xmm11[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm11 = xmm11[0],xmm4[0],xmm11[1],xmm4[1],xmm11[2],xmm4[2],xmm11[3],xmm4[3],xmm11[4],xmm4[4],xmm11[5],xmm4[5],xmm11[6],xmm4[6],xmm11[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm11, %xmm14
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm14 = xmm14[4],xmm4[4],xmm14[5],xmm4[5],xmm14[6],xmm4[6],xmm14[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm11 = xmm11[0],xmm4[0],xmm11[1],xmm4[1],xmm11[2],xmm4[2],xmm11[3],xmm4[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm4[0],xmm15[1],xmm4[1],xmm15[2],xmm4[2],xmm15[3],xmm4[3],xmm15[4],xmm4[4],xmm15[5],xmm4[5],xmm15[6],xmm4[6],xmm15[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm15, %xmm9
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm9 = xmm9[4],xmm4[4],xmm9[5],xmm4[5],xmm9[6],xmm4[6],xmm9[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm15 = xmm15[0],xmm4[0],xmm15[1],xmm4[1],xmm15[2],xmm4[2],xmm15[3],xmm4[3]
; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm0, %xmm7
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm7 = xmm7[4],xmm4[4],xmm7[5],xmm4[5],xmm7[6],xmm4[6],xmm7[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3],xmm3[4],xmm4[4],xmm3[5],xmm4[5],xmm3[6],xmm4[6],xmm3[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm3, %xmm6
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm4[4],xmm6[5],xmm4[5],xmm6[6],xmm4[6],xmm6[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1],xmm1[2],xmm4[2],xmm1[3],xmm4[3],xmm1[4],xmm4[4],xmm1[5],xmm4[5],xmm1[6],xmm4[6],xmm1[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm1, %xmm5
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm4[4],xmm5[5],xmm4[5],xmm5[6],xmm4[6],xmm5[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1],xmm1[2],xmm4[2],xmm1[3],xmm4[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1],xmm2[2],xmm4[2],xmm2[3],xmm4[3],xmm2[4],xmm4[4],xmm2[5],xmm4[5],xmm2[6],xmm4[6],xmm2[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm2, %xmm13
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm13 = xmm13[4],xmm4[4],xmm13[5],xmm4[5],xmm13[6],xmm4[6],xmm13[7],xmm4[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1],xmm2[2],xmm4[2],xmm2[3],xmm4[3]
; SSE2-NEXT:    paddd %xmm15, %xmm2
; SSE2-NEXT:    paddd %xmm9, %xmm13
; SSE2-NEXT:    paddd %xmm11, %xmm1
; SSE2-NEXT:    paddd %xmm14, %xmm5
; SSE2-NEXT:    paddd %xmm10, %xmm3
; SSE2-NEXT:    paddd %xmm12, %xmm6
; SSE2-NEXT:    paddd %xmm8, %xmm0
; SSE2-NEXT:    paddd -{{[0-9]+}}(%rsp), %xmm7 # 16-byte Folded Reload
; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [1,1,1,1]
; SSE2-NEXT:    paddd %xmm4, %xmm7
; SSE2-NEXT:    paddd %xmm4, %xmm0
; SSE2-NEXT:    paddd %xmm4, %xmm6
; SSE2-NEXT:    paddd %xmm4, %xmm3
; SSE2-NEXT:    paddd %xmm4, %xmm5
; SSE2-NEXT:    paddd %xmm4, %xmm1
; SSE2-NEXT:    paddd %xmm4, %xmm13
; SSE2-NEXT:    paddd %xmm4, %xmm2
; SSE2-NEXT:    psrld $1, %xmm0
; SSE2-NEXT:    psrld $1, %xmm7
; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
; SSE2-NEXT:    pand %xmm4, %xmm7
; SSE2-NEXT:    pand %xmm4, %xmm0
; SSE2-NEXT:    packuswb %xmm7, %xmm0
; SSE2-NEXT:    psrld $1, %xmm3
; SSE2-NEXT:    psrld $1, %xmm6
; SSE2-NEXT:    pand %xmm4, %xmm6
; SSE2-NEXT:    pand %xmm4, %xmm3
; SSE2-NEXT:    packuswb %xmm6, %xmm3
; SSE2-NEXT:    packuswb %xmm3, %xmm0
; SSE2-NEXT:    psrld $1, %xmm1
; SSE2-NEXT:    psrld $1, %xmm5
; SSE2-NEXT:    pand %xmm4, %xmm5
; SSE2-NEXT:    pand %xmm4, %xmm1
; SSE2-NEXT:    packuswb %xmm5, %xmm1
; SSE2-NEXT:    psrld $1, %xmm2
; SSE2-NEXT:    psrld $1, %xmm13
; SSE2-NEXT:    pand %xmm4, %xmm13
; SSE2-NEXT:    pand %xmm4, %xmm2
; SSE2-NEXT:    packuswb %xmm13, %xmm2
; SSE2-NEXT:    packuswb %xmm2, %xmm1
; SSE2-NEXT:    movdqu %xmm1, (%rax)
; SSE2-NEXT:    movdqu %xmm0, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v32i8_2:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovdqa (%rdi), %ymm0
; AVX2-NEXT:    vpavgb (%rsi), %ymm0, %ymm0
; AVX2-NEXT:    vmovdqu %ymm0, (%rax)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v32i8_2:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %ymm0
; AVX512F-NEXT:    vpavgb (%rsi), %ymm0, %ymm0
; AVX512F-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v32i8_2:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %ymm0
; AVX512BW-NEXT:    vpavgb (%rsi), %ymm0, %ymm0
; AVX512BW-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <32 x i8>, <32 x i8>* %a
  %2 = load <32 x i8>, <32 x i8>* %b
  %3 = zext <32 x i8> %1 to <32 x i32>
  %4 = zext <32 x i8> %2 to <32 x i32>
  %5 = add nuw nsw <32 x i32> %3, %4
  %6 = add nuw nsw <32 x i32> %5, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %7 = lshr <32 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <32 x i32> %7 to <32 x i8>
  store <32 x i8> %8, <32 x i8>* undef, align 4
  ret void
}

define void @avg_v64i8_2(<64 x i8>* %a, <64 x i8>* %b) {
; SSE2-LABEL: avg_v64i8_2:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rsi), %xmm15
; SSE2-NEXT:    movdqa 16(%rsi), %xmm13
; SSE2-NEXT:    movdqa 32(%rsi), %xmm2
; SSE2-NEXT:    movdqa 48(%rsi), %xmm3
; SSE2-NEXT:    pxor %xmm0, %xmm0
; SSE2-NEXT:    pshufd {{.*#+}} xmm7 = xmm15[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm0[0],xmm15[1],xmm0[1],xmm15[2],xmm0[2],xmm15[3],xmm0[3],xmm15[4],xmm0[4],xmm15[5],xmm0[5],xmm15[6],xmm0[6],xmm15[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm15, %xmm8
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm8 = xmm8[4],xmm0[4],xmm8[5],xmm0[5],xmm8[6],xmm0[6],xmm8[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm15 = xmm15[0],xmm0[0],xmm15[1],xmm0[1],xmm15[2],xmm0[2],xmm15[3],xmm0[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm0[0],xmm7[1],xmm0[1],xmm7[2],xmm0[2],xmm7[3],xmm0[3],xmm7[4],xmm0[4],xmm7[5],xmm0[5],xmm7[6],xmm0[6],xmm7[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm7, %xmm14
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm14 = xmm14[4],xmm0[4],xmm14[5],xmm0[5],xmm14[6],xmm0[6],xmm14[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm0[0],xmm7[1],xmm0[1],xmm7[2],xmm0[2],xmm7[3],xmm0[3]
; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm13[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm13 = xmm13[0],xmm0[0],xmm13[1],xmm0[1],xmm13[2],xmm0[2],xmm13[3],xmm0[3],xmm13[4],xmm0[4],xmm13[5],xmm0[5],xmm13[6],xmm0[6],xmm13[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm13, %xmm9
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm9 = xmm9[4],xmm0[4],xmm9[5],xmm0[5],xmm9[6],xmm0[6],xmm9[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm13 = xmm13[0],xmm0[0],xmm13[1],xmm0[1],xmm13[2],xmm0[2],xmm13[3],xmm0[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1],xmm6[2],xmm0[2],xmm6[3],xmm0[3],xmm6[4],xmm0[4],xmm6[5],xmm0[5],xmm6[6],xmm0[6],xmm6[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm6, %xmm12
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm12 = xmm12[4],xmm0[4],xmm12[5],xmm0[5],xmm12[6],xmm0[6],xmm12[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1],xmm6[2],xmm0[2],xmm6[3],xmm0[3]
; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm2[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm2, %xmm11
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm11 = xmm11[4],xmm0[4],xmm11[5],xmm0[5],xmm11[6],xmm0[6],xmm11[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1],xmm5[2],xmm0[2],xmm5[3],xmm0[3],xmm5[4],xmm0[4],xmm5[5],xmm0[5],xmm5[6],xmm0[6],xmm5[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm5, %xmm10
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm10 = xmm10[4],xmm0[4],xmm10[5],xmm0[5],xmm10[6],xmm0[6],xmm10[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1],xmm5[2],xmm0[2],xmm5[3],xmm0[3]
; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm3, %xmm4
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm0[4],xmm4[5],xmm0[5],xmm4[6],xmm0[6],xmm4[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm4, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm1, %xmm4
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm0[4],xmm4[5],xmm0[5],xmm4[6],xmm0[6],xmm4[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; SSE2-NEXT:    paddd %xmm1, %xmm1
; SSE2-NEXT:    paddd %xmm4, %xmm4
; SSE2-NEXT:    movdqa %xmm4, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    paddd %xmm3, %xmm3
; SSE2-NEXT:    movdqa -{{[0-9]+}}(%rsp), %xmm4 # 16-byte Reload
; SSE2-NEXT:    paddd %xmm4, %xmm4
; SSE2-NEXT:    paddd %xmm5, %xmm5
; SSE2-NEXT:    paddd %xmm10, %xmm10
; SSE2-NEXT:    paddd %xmm2, %xmm2
; SSE2-NEXT:    paddd %xmm11, %xmm11
; SSE2-NEXT:    paddd %xmm6, %xmm6
; SSE2-NEXT:    paddd %xmm12, %xmm12
; SSE2-NEXT:    paddd %xmm13, %xmm13
; SSE2-NEXT:    paddd %xmm9, %xmm9
; SSE2-NEXT:    paddd %xmm7, %xmm7
; SSE2-NEXT:    paddd %xmm14, %xmm14
; SSE2-NEXT:    paddd %xmm15, %xmm15
; SSE2-NEXT:    paddd %xmm8, %xmm8
; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [1,1,1,1]
; SSE2-NEXT:    paddd %xmm0, %xmm8
; SSE2-NEXT:    paddd %xmm0, %xmm15
; SSE2-NEXT:    paddd %xmm0, %xmm14
; SSE2-NEXT:    paddd %xmm0, %xmm7
; SSE2-NEXT:    paddd %xmm0, %xmm9
; SSE2-NEXT:    paddd %xmm0, %xmm13
; SSE2-NEXT:    paddd %xmm0, %xmm12
; SSE2-NEXT:    paddd %xmm0, %xmm6
; SSE2-NEXT:    paddd %xmm0, %xmm11
; SSE2-NEXT:    paddd %xmm0, %xmm2
; SSE2-NEXT:    paddd %xmm0, %xmm10
; SSE2-NEXT:    paddd %xmm0, %xmm5
; SSE2-NEXT:    paddd %xmm0, %xmm4
; SSE2-NEXT:    movdqa %xmm4, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    paddd %xmm0, %xmm3
; SSE2-NEXT:    movdqa -{{[0-9]+}}(%rsp), %xmm4 # 16-byte Reload
; SSE2-NEXT:    paddd %xmm0, %xmm4
; SSE2-NEXT:    paddd %xmm0, %xmm1
; SSE2-NEXT:    psrld $1, %xmm15
; SSE2-NEXT:    psrld $1, %xmm8
; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
; SSE2-NEXT:    pand %xmm0, %xmm8
; SSE2-NEXT:    pand %xmm0, %xmm15
; SSE2-NEXT:    packuswb %xmm8, %xmm15
; SSE2-NEXT:    psrld $1, %xmm7
; SSE2-NEXT:    psrld $1, %xmm14
; SSE2-NEXT:    pand %xmm0, %xmm14
; SSE2-NEXT:    pand %xmm0, %xmm7
; SSE2-NEXT:    packuswb %xmm14, %xmm7
; SSE2-NEXT:    packuswb %xmm7, %xmm15
; SSE2-NEXT:    psrld $1, %xmm13
; SSE2-NEXT:    psrld $1, %xmm9
; SSE2-NEXT:    pand %xmm0, %xmm9
; SSE2-NEXT:    pand %xmm0, %xmm13
; SSE2-NEXT:    packuswb %xmm9, %xmm13
; SSE2-NEXT:    psrld $1, %xmm6
; SSE2-NEXT:    psrld $1, %xmm12
; SSE2-NEXT:    pand %xmm0, %xmm12
; SSE2-NEXT:    pand %xmm0, %xmm6
; SSE2-NEXT:    packuswb %xmm12, %xmm6
; SSE2-NEXT:    packuswb %xmm6, %xmm13
; SSE2-NEXT:    psrld $1, %xmm2
; SSE2-NEXT:    psrld $1, %xmm11
; SSE2-NEXT:    pand %xmm0, %xmm11
; SSE2-NEXT:    pand %xmm0, %xmm2
; SSE2-NEXT:    packuswb %xmm11, %xmm2
; SSE2-NEXT:    psrld $1, %xmm5
; SSE2-NEXT:    psrld $1, %xmm10
; SSE2-NEXT:    pand %xmm0, %xmm10
; SSE2-NEXT:    pand %xmm0, %xmm5
; SSE2-NEXT:    packuswb %xmm10, %xmm5
; SSE2-NEXT:    packuswb %xmm5, %xmm2
; SSE2-NEXT:    psrld $1, %xmm3
; SSE2-NEXT:    movdqa -{{[0-9]+}}(%rsp), %xmm5 # 16-byte Reload
; SSE2-NEXT:    psrld $1, %xmm5
; SSE2-NEXT:    pand %xmm0, %xmm5
; SSE2-NEXT:    pand %xmm0, %xmm3
; SSE2-NEXT:    packuswb %xmm5, %xmm3
; SSE2-NEXT:    psrld $1, %xmm1
; SSE2-NEXT:    movdqa %xmm4, %xmm5
; SSE2-NEXT:    psrld $1, %xmm5
; SSE2-NEXT:    pand %xmm0, %xmm5
; SSE2-NEXT:    pand %xmm0, %xmm1
; SSE2-NEXT:    packuswb %xmm5, %xmm1
; SSE2-NEXT:    packuswb %xmm1, %xmm3
; SSE2-NEXT:    movdqu %xmm3, (%rax)
; SSE2-NEXT:    movdqu %xmm2, (%rax)
; SSE2-NEXT:    movdqu %xmm13, (%rax)
; SSE2-NEXT:    movdqu %xmm15, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v64i8_2:
; AVX2:       # BB#0:
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm5 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm6 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpaddd %ymm7, %ymm7, %ymm7
; AVX2-NEXT:    vpaddd %ymm6, %ymm6, %ymm6
; AVX2-NEXT:    vpaddd %ymm5, %ymm5, %ymm5
; AVX2-NEXT:    vpaddd %ymm4, %ymm4, %ymm4
; AVX2-NEXT:    vpaddd %ymm3, %ymm3, %ymm3
; AVX2-NEXT:    vpaddd %ymm2, %ymm2, %ymm2
; AVX2-NEXT:    vpaddd %ymm1, %ymm1, %ymm1
; AVX2-NEXT:    vpaddd %ymm0, %ymm0, %ymm0
; AVX2-NEXT:    vpbroadcastd {{.*}}(%rip), %ymm8
; AVX2-NEXT:    vpaddd %ymm8, %ymm0, %ymm9
; AVX2-NEXT:    vpaddd %ymm8, %ymm1, %ymm10
; AVX2-NEXT:    vpaddd %ymm8, %ymm2, %ymm2
; AVX2-NEXT:    vpaddd %ymm8, %ymm3, %ymm3
; AVX2-NEXT:    vpaddd %ymm8, %ymm4, %ymm4
; AVX2-NEXT:    vpaddd %ymm8, %ymm5, %ymm5
; AVX2-NEXT:    vpaddd %ymm8, %ymm6, %ymm1
; AVX2-NEXT:    vpaddd %ymm8, %ymm7, %ymm0
; AVX2-NEXT:    vpsrld $1, %ymm0, %ymm11
; AVX2-NEXT:    vpsrld $1, %ymm1, %ymm12
; AVX2-NEXT:    vpsrld $1, %ymm5, %ymm5
; AVX2-NEXT:    vpsrld $1, %ymm4, %ymm4
; AVX2-NEXT:    vpsrld $1, %ymm3, %ymm6
; AVX2-NEXT:    vpsrld $1, %ymm2, %ymm7
; AVX2-NEXT:    vpsrld $1, %ymm10, %ymm8
; AVX2-NEXT:    vpsrld $1, %ymm9, %ymm3
; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,1,4,5,8,9,12,13,128,128,128,128,128,128,128,128,0,1,4,5,8,9,12,13,128,128,128,128,128,128,128,128]
; AVX2-NEXT:    vpshufb %ymm2, %ymm3, %ymm3
; AVX2-NEXT:    vpermq {{.*#+}} ymm9 = ymm3[0,2,2,3]
; AVX2-NEXT:    vmovdqa {{.*#+}} xmm3 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
; AVX2-NEXT:    vpshufb %xmm3, %xmm9, %xmm0
; AVX2-NEXT:    vpshufb %ymm2, %ymm8, %ymm8
; AVX2-NEXT:    vpermq {{.*#+}} ymm8 = ymm8[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm8, %xmm1
; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; AVX2-NEXT:    vpshufb %ymm2, %ymm7, %ymm1
; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
; AVX2-NEXT:    vpshufb %ymm2, %ymm6, %ymm6
; AVX2-NEXT:    vpermq {{.*#+}} ymm6 = ymm6[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm6, %xmm6
; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm6[0],xmm1[0]
; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
; AVX2-NEXT:    vpshufb %ymm2, %ymm4, %ymm1
; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
; AVX2-NEXT:    vpshufb %ymm2, %ymm5, %ymm4
; AVX2-NEXT:    vpermq {{.*#+}} ymm4 = ymm4[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm4, %xmm4
; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm4[0],xmm1[0]
; AVX2-NEXT:    vpshufb %ymm2, %ymm12, %ymm4
; AVX2-NEXT:    vpermq {{.*#+}} ymm4 = ymm4[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm4, %xmm4
; AVX2-NEXT:    vpshufb %ymm2, %ymm11, %ymm2
; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm2, %xmm2
; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm4[0]
; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm2, %ymm1
; AVX2-NEXT:    vmovdqu %ymm1, (%rax)
; AVX2-NEXT:    vmovdqu %ymm0, (%rax)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v64i8_2:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpaddd %zmm3, %zmm3, %zmm3
; AVX512F-NEXT:    vpaddd %zmm2, %zmm2, %zmm2
; AVX512F-NEXT:    vpaddd %zmm1, %zmm1, %zmm1
; AVX512F-NEXT:    vpaddd %zmm0, %zmm0, %zmm0
; AVX512F-NEXT:    vpbroadcastd {{.*}}(%rip), %zmm4
; AVX512F-NEXT:    vpaddd %zmm4, %zmm0, %zmm0
; AVX512F-NEXT:    vpaddd %zmm4, %zmm1, %zmm1
; AVX512F-NEXT:    vpaddd %zmm4, %zmm2, %zmm2
; AVX512F-NEXT:    vpaddd %zmm4, %zmm3, %zmm3
; AVX512F-NEXT:    vpsrld $1, %zmm3, %zmm3
; AVX512F-NEXT:    vpsrld $1, %zmm2, %zmm2
; AVX512F-NEXT:    vpsrld $1, %zmm1, %zmm1
; AVX512F-NEXT:    vpsrld $1, %zmm0, %zmm0
; AVX512F-NEXT:    vpmovdb %zmm0, %xmm0
; AVX512F-NEXT:    vpmovdb %zmm1, %xmm1
; AVX512F-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX512F-NEXT:    vpmovdb %zmm2, %xmm1
; AVX512F-NEXT:    vpmovdb %zmm3, %xmm2
; AVX512F-NEXT:    vinserti128 $1, %xmm2, %ymm1, %ymm1
; AVX512F-NEXT:    vmovdqu %ymm1, (%rax)
; AVX512F-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v64i8_2:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqu8 (%rsi), %zmm0
; AVX512BW-NEXT:    vpavgb %zmm0, %zmm0, %zmm0
; AVX512BW-NEXT:    vmovdqu8 %zmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <64 x i8>, <64 x i8>* %a
  %2 = load <64 x i8>, <64 x i8>* %b
  %3 = zext <64 x i8> %1 to <64 x i32>
  %4 = zext <64 x i8> %2 to <64 x i32>
  %5 = add nuw nsw <64 x i32> %4, %4
  %6 = add nuw nsw <64 x i32> %5, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %7 = lshr <64 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <64 x i32> %7 to <64 x i8>
  store <64 x i8> %8, <64 x i8>* undef, align 4
  ret void
}


define void @avg_v4i16_2(<4 x i16>* %a, <4 x i16>* %b) {
; SSE2-LABEL: avg_v4i16_2:
; SSE2:       # BB#0:
; SSE2-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
; SSE2-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
; SSE2-NEXT:    pavgw %xmm0, %xmm1
; SSE2-NEXT:    movq %xmm1, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v4i16_2:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
; AVX2-NEXT:    vpavgw %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vmovq %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v4i16_2:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512F-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
; AVX512F-NEXT:    vpavgw %xmm1, %xmm0, %xmm0
; AVX512F-NEXT:    vmovq %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v4i16_2:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512BW-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
; AVX512BW-NEXT:    vpavgw %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT:    vmovq %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <4 x i16>, <4 x i16>* %a
  %2 = load <4 x i16>, <4 x i16>* %b
  %3 = zext <4 x i16> %1 to <4 x i32>
  %4 = zext <4 x i16> %2 to <4 x i32>
  %5 = add nuw nsw <4 x i32> %3, %4
  %6 = add nuw nsw <4 x i32> %5, <i32 1, i32 1, i32 1, i32 1>
  %7 = lshr <4 x i32> %6, <i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <4 x i32> %7 to <4 x i16>
  store <4 x i16> %8, <4 x i16>* undef, align 4
  ret void
}

define void @avg_v8i16_2(<8 x i16>* %a, <8 x i16>* %b) {
; SSE2-LABEL: avg_v8i16_2:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pavgw (%rsi), %xmm0
; SSE2-NEXT:    movdqu %xmm0, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v8i16_2:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovdqa (%rdi), %xmm0
; AVX2-NEXT:    vpavgw (%rsi), %xmm0, %xmm0
; AVX2-NEXT:    vmovdqu %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v8i16_2:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpavgw (%rsi), %xmm0, %xmm0
; AVX512F-NEXT:    vmovdqu %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v8i16_2:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpavgw (%rsi), %xmm0, %xmm0
; AVX512BW-NEXT:    vmovdqu %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <8 x i16>, <8 x i16>* %a
  %2 = load <8 x i16>, <8 x i16>* %b
  %3 = zext <8 x i16> %1 to <8 x i32>
  %4 = zext <8 x i16> %2 to <8 x i32>
  %5 = add nuw nsw <8 x i32> %3, %4
  %6 = add nuw nsw <8 x i32> %5, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %7 = lshr <8 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <8 x i32> %7 to <8 x i16>
  store <8 x i16> %8, <8 x i16>* undef, align 4
  ret void
}

define void @avg_v16i16_2(<16 x i16>* %a, <16 x i16>* %b) {
; SSE2-LABEL: avg_v16i16_2:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm4
; SSE2-NEXT:    movdqa 16(%rdi), %xmm5
; SSE2-NEXT:    movdqa (%rsi), %xmm0
; SSE2-NEXT:    movdqa 16(%rsi), %xmm1
; SSE2-NEXT:    pxor %xmm6, %xmm6
; SSE2-NEXT:    movdqa %xmm4, %xmm8
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm8 = xmm8[4],xmm6[4],xmm8[5],xmm6[5],xmm8[6],xmm6[6],xmm8[7],xmm6[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm6[0],xmm4[1],xmm6[1],xmm4[2],xmm6[2],xmm4[3],xmm6[3]
; SSE2-NEXT:    movdqa %xmm5, %xmm7
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm7 = xmm7[4],xmm6[4],xmm7[5],xmm6[5],xmm7[6],xmm6[6],xmm7[7],xmm6[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm6[0],xmm5[1],xmm6[1],xmm5[2],xmm6[2],xmm5[3],xmm6[3]
; SSE2-NEXT:    movdqa %xmm0, %xmm3
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm6[4],xmm3[5],xmm6[5],xmm3[6],xmm6[6],xmm3[7],xmm6[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm6[0],xmm0[1],xmm6[1],xmm0[2],xmm6[2],xmm0[3],xmm6[3]
; SSE2-NEXT:    movdqa %xmm1, %xmm2
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm6[4],xmm2[5],xmm6[5],xmm2[6],xmm6[6],xmm2[7],xmm6[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm6[0],xmm1[1],xmm6[1],xmm1[2],xmm6[2],xmm1[3],xmm6[3]
; SSE2-NEXT:    paddd %xmm5, %xmm1
; SSE2-NEXT:    paddd %xmm7, %xmm2
; SSE2-NEXT:    paddd %xmm4, %xmm0
; SSE2-NEXT:    paddd %xmm8, %xmm3
; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [1,1,1,1]
; SSE2-NEXT:    paddd %xmm4, %xmm3
; SSE2-NEXT:    paddd %xmm4, %xmm0
; SSE2-NEXT:    paddd %xmm4, %xmm2
; SSE2-NEXT:    paddd %xmm4, %xmm1
; SSE2-NEXT:    psrld $1, %xmm1
; SSE2-NEXT:    psrld $1, %xmm2
; SSE2-NEXT:    psrld $1, %xmm0
; SSE2-NEXT:    psrld $1, %xmm3
; SSE2-NEXT:    pslld $16, %xmm3
; SSE2-NEXT:    psrad $16, %xmm3
; SSE2-NEXT:    pslld $16, %xmm0
; SSE2-NEXT:    psrad $16, %xmm0
; SSE2-NEXT:    packssdw %xmm3, %xmm0
; SSE2-NEXT:    pslld $16, %xmm2
; SSE2-NEXT:    psrad $16, %xmm2
; SSE2-NEXT:    pslld $16, %xmm1
; SSE2-NEXT:    psrad $16, %xmm1
; SSE2-NEXT:    packssdw %xmm2, %xmm1
; SSE2-NEXT:    movdqu %xmm1, (%rax)
; SSE2-NEXT:    movdqu %xmm0, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v16i16_2:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovdqa (%rdi), %ymm0
; AVX2-NEXT:    vpavgw (%rsi), %ymm0, %ymm0
; AVX2-NEXT:    vmovdqu %ymm0, (%rax)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v16i16_2:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %ymm0
; AVX512F-NEXT:    vpavgw (%rsi), %ymm0, %ymm0
; AVX512F-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v16i16_2:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %ymm0
; AVX512BW-NEXT:    vpavgw (%rsi), %ymm0, %ymm0
; AVX512BW-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <16 x i16>, <16 x i16>* %a
  %2 = load <16 x i16>, <16 x i16>* %b
  %3 = zext <16 x i16> %1 to <16 x i32>
  %4 = zext <16 x i16> %2 to <16 x i32>
  %5 = add nuw nsw <16 x i32> %3, %4
  %6 = add nuw nsw <16 x i32> %5, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %7 = lshr <16 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <16 x i32> %7 to <16 x i16>
  store <16 x i16> %8, <16 x i16>* undef, align 4
  ret void
}

define void @avg_v32i16_2(<32 x i16>* %a, <32 x i16>* %b) {
; SSE2-LABEL: avg_v32i16_2:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm10
; SSE2-NEXT:    movdqa 16(%rdi), %xmm9
; SSE2-NEXT:    movdqa 32(%rdi), %xmm11
; SSE2-NEXT:    movdqa 48(%rdi), %xmm8
; SSE2-NEXT:    movdqa (%rsi), %xmm14
; SSE2-NEXT:    movdqa 16(%rsi), %xmm1
; SSE2-NEXT:    movdqa 32(%rsi), %xmm2
; SSE2-NEXT:    movdqa 48(%rsi), %xmm3
; SSE2-NEXT:    pxor %xmm0, %xmm0
; SSE2-NEXT:    movdqa %xmm10, %xmm4
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm0[4],xmm4[5],xmm0[5],xmm4[6],xmm0[6],xmm4[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm4, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm10 = xmm10[0],xmm0[0],xmm10[1],xmm0[1],xmm10[2],xmm0[2],xmm10[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm9, %xmm12
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm12 = xmm12[4],xmm0[4],xmm12[5],xmm0[5],xmm12[6],xmm0[6],xmm12[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm9 = xmm9[0],xmm0[0],xmm9[1],xmm0[1],xmm9[2],xmm0[2],xmm9[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm11, %xmm15
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm15 = xmm15[4],xmm0[4],xmm15[5],xmm0[5],xmm15[6],xmm0[6],xmm15[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm11 = xmm11[0],xmm0[0],xmm11[1],xmm0[1],xmm11[2],xmm0[2],xmm11[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm8, %xmm13
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm13 = xmm13[4],xmm0[4],xmm13[5],xmm0[5],xmm13[6],xmm0[6],xmm13[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm0[0],xmm8[1],xmm0[1],xmm8[2],xmm0[2],xmm8[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm14, %xmm7
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm7 = xmm7[4],xmm0[4],xmm7[5],xmm0[5],xmm7[6],xmm0[6],xmm7[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm14 = xmm14[0],xmm0[0],xmm14[1],xmm0[1],xmm14[2],xmm0[2],xmm14[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm1, %xmm6
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm0[4],xmm6[5],xmm0[5],xmm6[6],xmm0[6],xmm6[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm2, %xmm5
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm0[4],xmm5[5],xmm0[5],xmm5[6],xmm0[6],xmm5[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm3, %xmm4
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm0[4],xmm4[5],xmm0[5],xmm4[6],xmm0[6],xmm4[7],xmm0[7]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
; SSE2-NEXT:    paddd %xmm8, %xmm3
; SSE2-NEXT:    paddd %xmm13, %xmm4
; SSE2-NEXT:    paddd %xmm11, %xmm2
; SSE2-NEXT:    paddd %xmm15, %xmm5
; SSE2-NEXT:    paddd %xmm9, %xmm1
; SSE2-NEXT:    paddd %xmm12, %xmm6
; SSE2-NEXT:    paddd %xmm10, %xmm14
; SSE2-NEXT:    paddd -{{[0-9]+}}(%rsp), %xmm7 # 16-byte Folded Reload
; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [1,1,1,1]
; SSE2-NEXT:    paddd %xmm0, %xmm7
; SSE2-NEXT:    paddd %xmm0, %xmm14
; SSE2-NEXT:    paddd %xmm0, %xmm6
; SSE2-NEXT:    paddd %xmm0, %xmm1
; SSE2-NEXT:    paddd %xmm0, %xmm5
; SSE2-NEXT:    paddd %xmm0, %xmm2
; SSE2-NEXT:    paddd %xmm0, %xmm4
; SSE2-NEXT:    paddd %xmm0, %xmm3
; SSE2-NEXT:    psrld $1, %xmm14
; SSE2-NEXT:    psrld $1, %xmm7
; SSE2-NEXT:    pslld $16, %xmm7
; SSE2-NEXT:    psrad $16, %xmm7
; SSE2-NEXT:    pslld $16, %xmm14
; SSE2-NEXT:    psrad $16, %xmm14
; SSE2-NEXT:    packssdw %xmm7, %xmm14
; SSE2-NEXT:    psrld $1, %xmm1
; SSE2-NEXT:    psrld $1, %xmm6
; SSE2-NEXT:    pslld $16, %xmm6
; SSE2-NEXT:    psrad $16, %xmm6
; SSE2-NEXT:    pslld $16, %xmm1
; SSE2-NEXT:    psrad $16, %xmm1
; SSE2-NEXT:    packssdw %xmm6, %xmm1
; SSE2-NEXT:    psrld $1, %xmm2
; SSE2-NEXT:    psrld $1, %xmm5
; SSE2-NEXT:    pslld $16, %xmm5
; SSE2-NEXT:    psrad $16, %xmm5
; SSE2-NEXT:    pslld $16, %xmm2
; SSE2-NEXT:    psrad $16, %xmm2
; SSE2-NEXT:    packssdw %xmm5, %xmm2
; SSE2-NEXT:    psrld $1, %xmm3
; SSE2-NEXT:    psrld $1, %xmm4
; SSE2-NEXT:    pslld $16, %xmm4
; SSE2-NEXT:    psrad $16, %xmm4
; SSE2-NEXT:    pslld $16, %xmm3
; SSE2-NEXT:    psrad $16, %xmm3
; SSE2-NEXT:    packssdw %xmm4, %xmm3
; SSE2-NEXT:    movdqu %xmm3, (%rax)
; SSE2-NEXT:    movdqu %xmm2, (%rax)
; SSE2-NEXT:    movdqu %xmm1, (%rax)
; SSE2-NEXT:    movdqu %xmm14, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v32i16_2:
; AVX2:       # BB#0:
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm2 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm4 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm5 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm6 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpaddd %ymm7, %ymm3, %ymm3
; AVX2-NEXT:    vpaddd %ymm6, %ymm2, %ymm2
; AVX2-NEXT:    vpaddd %ymm5, %ymm1, %ymm1
; AVX2-NEXT:    vpaddd %ymm4, %ymm0, %ymm0
; AVX2-NEXT:    vpbroadcastd {{.*}}(%rip), %ymm4
; AVX2-NEXT:    vpaddd %ymm4, %ymm0, %ymm0
; AVX2-NEXT:    vpaddd %ymm4, %ymm1, %ymm1
; AVX2-NEXT:    vpaddd %ymm4, %ymm2, %ymm2
; AVX2-NEXT:    vpaddd %ymm4, %ymm3, %ymm3
; AVX2-NEXT:    vpsrld $1, %ymm3, %ymm3
; AVX2-NEXT:    vpsrld $1, %ymm2, %ymm2
; AVX2-NEXT:    vpsrld $1, %ymm1, %ymm1
; AVX2-NEXT:    vpsrld $1, %ymm0, %ymm0
; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = [0,1,4,5,8,9,12,13,128,128,128,128,128,128,128,128,0,1,4,5,8,9,12,13,128,128,128,128,128,128,128,128]
; AVX2-NEXT:    vpshufb %ymm4, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
; AVX2-NEXT:    vpshufb %ymm4, %ymm1, %ymm1
; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
; AVX2-NEXT:    vpshufb %ymm4, %ymm2, %ymm1
; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
; AVX2-NEXT:    vpshufb %ymm4, %ymm3, %ymm2
; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
; AVX2-NEXT:    vinserti128 $1, %xmm2, %ymm1, %ymm1
; AVX2-NEXT:    vmovdqu %ymm1, (%rax)
; AVX2-NEXT:    vmovdqu %ymm0, (%rax)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v32i16_2:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vpmovzxwd {{.*#+}} zmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero
; AVX512F-NEXT:    vpmovzxwd {{.*#+}} zmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero
; AVX512F-NEXT:    vpmovzxwd {{.*#+}} zmm2 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero
; AVX512F-NEXT:    vpmovzxwd {{.*#+}} zmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero
; AVX512F-NEXT:    vpaddd %zmm3, %zmm1, %zmm1
; AVX512F-NEXT:    vpaddd %zmm2, %zmm0, %zmm0
; AVX512F-NEXT:    vpbroadcastd {{.*}}(%rip), %zmm2
; AVX512F-NEXT:    vpaddd %zmm2, %zmm0, %zmm0
; AVX512F-NEXT:    vpaddd %zmm2, %zmm1, %zmm1
; AVX512F-NEXT:    vpsrld $1, %zmm1, %zmm1
; AVX512F-NEXT:    vpsrld $1, %zmm0, %zmm0
; AVX512F-NEXT:    vpmovdw %zmm0, (%rax)
; AVX512F-NEXT:    vpmovdw %zmm1, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v32i16_2:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqu16 (%rdi), %zmm0
; AVX512BW-NEXT:    vpavgw (%rsi), %zmm0, %zmm0
; AVX512BW-NEXT:    vmovdqu16 %zmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <32 x i16>, <32 x i16>* %a
  %2 = load <32 x i16>, <32 x i16>* %b
  %3 = zext <32 x i16> %1 to <32 x i32>
  %4 = zext <32 x i16> %2 to <32 x i32>
  %5 = add nuw nsw <32 x i32> %3, %4
  %6 = add nuw nsw <32 x i32> %5, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %7 = lshr <32 x i32> %6, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %8 = trunc <32 x i32> %7 to <32 x i16>
  store <32 x i16> %8, <32 x i16>* undef, align 4
  ret void
}

define void @avg_v4i8_const(<4 x i8>* %a) {
; SSE2-LABEL: avg_v4i8_const:
; SSE2:       # BB#0:
; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; SSE2-NEXT:    pavgb {{.*}}(%rip), %xmm0
; SSE2-NEXT:    movd %xmm0, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v4i8_const:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX2-NEXT:    vpavgb {{.*}}(%rip), %xmm0, %xmm0
; AVX2-NEXT:    vmovd %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v4i8_const:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX512F-NEXT:    vpavgb {{.*}}(%rip), %xmm0, %xmm0
; AVX512F-NEXT:    vmovd %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v4i8_const:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX512BW-NEXT:    vpavgb {{.*}}(%rip), %xmm0, %xmm0
; AVX512BW-NEXT:    vmovd %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <4 x i8>, <4 x i8>* %a
  %2 = zext <4 x i8> %1 to <4 x i32>
  %3 = add nuw nsw <4 x i32> %2, <i32 1, i32 2, i32 3, i32 4>
  %4 = lshr <4 x i32> %3, <i32 1, i32 1, i32 1, i32 1>
  %5 = trunc <4 x i32> %4 to <4 x i8>
  store <4 x i8> %5, <4 x i8>* undef, align 4
  ret void
}

define void @avg_v8i8_const(<8 x i8>* %a) {
; SSE2-LABEL: avg_v8i8_const:
; SSE2:       # BB#0:
; SSE2-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
; SSE2-NEXT:    pavgb {{.*}}(%rip), %xmm0
; SSE2-NEXT:    movq %xmm0, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v8i8_const:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX2-NEXT:    vpavgb {{.*}}(%rip), %xmm0, %xmm0
; AVX2-NEXT:    vmovq %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v8i8_const:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512F-NEXT:    vpavgb {{.*}}(%rip), %xmm0, %xmm0
; AVX512F-NEXT:    vmovq %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v8i8_const:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512BW-NEXT:    vpavgb {{.*}}(%rip), %xmm0, %xmm0
; AVX512BW-NEXT:    vmovq %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <8 x i8>, <8 x i8>* %a
  %2 = zext <8 x i8> %1 to <8 x i32>
  %3 = add nuw nsw <8 x i32> %2, <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
  %4 = lshr <8 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %5 = trunc <8 x i32> %4 to <8 x i8>
  store <8 x i8> %5, <8 x i8>* undef, align 4
  ret void
}

define void @avg_v16i8_const(<16 x i8>* %a) {
; SSE2-LABEL: avg_v16i8_const:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pavgb {{.*}}(%rip), %xmm0
; SSE2-NEXT:    movdqu %xmm0, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v16i8_const:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovdqa (%rdi), %xmm0
; AVX2-NEXT:    vpavgb {{.*}}(%rip), %xmm0, %xmm0
; AVX2-NEXT:    vmovdqu %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v16i8_const:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpavgb {{.*}}(%rip), %xmm0, %xmm0
; AVX512F-NEXT:    vmovdqu %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v16i8_const:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpavgb {{.*}}(%rip), %xmm0, %xmm0
; AVX512BW-NEXT:    vmovdqu %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <16 x i8>, <16 x i8>* %a
  %2 = zext <16 x i8> %1 to <16 x i32>
  %3 = add nuw nsw <16 x i32> %2, <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
  %4 = lshr <16 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %5 = trunc <16 x i32> %4 to <16 x i8>
  store <16 x i8> %5, <16 x i8>* undef, align 4
  ret void
}

define void @avg_v32i8_const(<32 x i8>* %a) {
; SSE2-LABEL: avg_v32i8_const:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm4
; SSE2-NEXT:    movdqa 16(%rdi), %xmm2
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[2,3,0,1]
; SSE2-NEXT:    pxor %xmm1, %xmm1
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; SSE2-NEXT:    movdqa %xmm0, %xmm8
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm1[0],xmm8[1],xmm1[1],xmm8[2],xmm1[2],xmm8[3],xmm1[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
; SSE2-NEXT:    movdqa %xmm2, %xmm3
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm4[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm1[0],xmm6[1],xmm1[1],xmm6[2],xmm1[2],xmm6[3],xmm1[3],xmm6[4],xmm1[4],xmm6[5],xmm1[5],xmm6[6],xmm1[6],xmm6[7],xmm1[7]
; SSE2-NEXT:    movdqa %xmm6, %xmm7
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm1[0],xmm7[1],xmm1[1],xmm7[2],xmm1[2],xmm7[3],xmm1[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm1[4],xmm6[5],xmm1[5],xmm6[6],xmm1[6],xmm6[7],xmm1[7]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1],xmm4[2],xmm1[2],xmm4[3],xmm1[3],xmm4[4],xmm1[4],xmm4[5],xmm1[5],xmm4[6],xmm1[6],xmm4[7],xmm1[7]
; SSE2-NEXT:    movdqa %xmm4, %xmm5
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm1[0],xmm5[1],xmm1[1],xmm5[2],xmm1[2],xmm5[3],xmm1[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm1[4],xmm4[5],xmm1[5],xmm4[6],xmm1[6],xmm4[7],xmm1[7]
; SSE2-NEXT:    movdqa {{.*#+}} xmm9 = [5,6,7,8]
; SSE2-NEXT:    paddd %xmm9, %xmm4
; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [1,2,3,4]
; SSE2-NEXT:    paddd %xmm1, %xmm5
; SSE2-NEXT:    paddd %xmm9, %xmm6
; SSE2-NEXT:    paddd %xmm1, %xmm7
; SSE2-NEXT:    paddd %xmm9, %xmm2
; SSE2-NEXT:    paddd %xmm1, %xmm3
; SSE2-NEXT:    paddd %xmm9, %xmm0
; SSE2-NEXT:    paddd %xmm1, %xmm8
; SSE2-NEXT:    psrld $1, %xmm8
; SSE2-NEXT:    psrld $1, %xmm0
; SSE2-NEXT:    psrld $1, %xmm3
; SSE2-NEXT:    psrld $1, %xmm2
; SSE2-NEXT:    psrld $1, %xmm7
; SSE2-NEXT:    psrld $1, %xmm6
; SSE2-NEXT:    psrld $1, %xmm5
; SSE2-NEXT:    psrld $1, %xmm4
; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
; SSE2-NEXT:    pand %xmm1, %xmm4
; SSE2-NEXT:    pand %xmm1, %xmm5
; SSE2-NEXT:    packuswb %xmm4, %xmm5
; SSE2-NEXT:    pand %xmm1, %xmm6
; SSE2-NEXT:    pand %xmm1, %xmm7
; SSE2-NEXT:    packuswb %xmm6, %xmm7
; SSE2-NEXT:    packuswb %xmm7, %xmm5
; SSE2-NEXT:    pand %xmm1, %xmm2
; SSE2-NEXT:    pand %xmm1, %xmm3
; SSE2-NEXT:    packuswb %xmm2, %xmm3
; SSE2-NEXT:    pand %xmm1, %xmm0
; SSE2-NEXT:    pand %xmm1, %xmm8
; SSE2-NEXT:    packuswb %xmm0, %xmm8
; SSE2-NEXT:    packuswb %xmm8, %xmm3
; SSE2-NEXT:    movdqu %xmm3, (%rax)
; SSE2-NEXT:    movdqu %xmm5, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v32i8_const:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovdqa (%rdi), %ymm0
; AVX2-NEXT:    vpavgb {{.*}}(%rip), %ymm0, %ymm0
; AVX2-NEXT:    vmovdqu %ymm0, (%rax)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v32i8_const:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %ymm0
; AVX512F-NEXT:    vpavgb {{.*}}(%rip), %ymm0, %ymm0
; AVX512F-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v32i8_const:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %ymm0
; AVX512BW-NEXT:    vpavgb {{.*}}(%rip), %ymm0, %ymm0
; AVX512BW-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <32 x i8>, <32 x i8>* %a
  %2 = zext <32 x i8> %1 to <32 x i32>
  %3 = add nuw nsw <32 x i32> %2, <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
  %4 = lshr <32 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %5 = trunc <32 x i32> %4 to <32 x i8>
  store <32 x i8> %5, <32 x i8>* undef, align 4
  ret void
}

define void @avg_v64i8_const(<64 x i8>* %a) {
; SSE2-LABEL: avg_v64i8_const:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm7
; SSE2-NEXT:    movdqa 16(%rdi), %xmm1
; SSE2-NEXT:    movdqa 32(%rdi), %xmm14
; SSE2-NEXT:    movdqa 48(%rdi), %xmm11
; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm11[2,3,0,1]
; SSE2-NEXT:    pxor %xmm0, %xmm0
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm3, %xmm2
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
; SSE2-NEXT:    movdqa %xmm2, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm3, %xmm9
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm11 = xmm11[0],xmm0[0],xmm11[1],xmm0[1],xmm11[2],xmm0[2],xmm11[3],xmm0[3],xmm11[4],xmm0[4],xmm11[5],xmm0[5],xmm11[6],xmm0[6],xmm11[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm11, %xmm12
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm12 = xmm12[0],xmm0[0],xmm12[1],xmm0[1],xmm12[2],xmm0[2],xmm12[3],xmm0[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm11 = xmm11[4],xmm0[4],xmm11[5],xmm0[5],xmm11[6],xmm0[6],xmm11[7],xmm0[7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm13 = xmm14[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm13 = xmm13[0],xmm0[0],xmm13[1],xmm0[1],xmm13[2],xmm0[2],xmm13[3],xmm0[3],xmm13[4],xmm0[4],xmm13[5],xmm0[5],xmm13[6],xmm0[6],xmm13[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm13, %xmm10
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm10 = xmm10[0],xmm0[0],xmm10[1],xmm0[1],xmm10[2],xmm0[2],xmm10[3],xmm0[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm13 = xmm13[4],xmm0[4],xmm13[5],xmm0[5],xmm13[6],xmm0[6],xmm13[7],xmm0[7]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm14 = xmm14[0],xmm0[0],xmm14[1],xmm0[1],xmm14[2],xmm0[2],xmm14[3],xmm0[3],xmm14[4],xmm0[4],xmm14[5],xmm0[5],xmm14[6],xmm0[6],xmm14[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm14, %xmm15
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm15 = xmm15[0],xmm0[0],xmm15[1],xmm0[1],xmm15[2],xmm0[2],xmm15[3],xmm0[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm14 = xmm14[4],xmm0[4],xmm14[5],xmm0[5],xmm14[6],xmm0[6],xmm14[7],xmm0[7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm3, %xmm6
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1],xmm6[2],xmm0[2],xmm6[3],xmm0[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm1, %xmm8
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm0[0],xmm8[1],xmm0[1],xmm8[2],xmm0[2],xmm8[3],xmm0[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm1 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm7[2,3,0,1]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1],xmm4[2],xmm0[2],xmm4[3],xmm0[3],xmm4[4],xmm0[4],xmm4[5],xmm0[5],xmm4[6],xmm0[6],xmm4[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm4, %xmm5
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1],xmm5[2],xmm0[2],xmm5[3],xmm0[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm0[4],xmm4[5],xmm0[5],xmm4[6],xmm0[6],xmm4[7],xmm0[7]
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm0[0],xmm7[1],xmm0[1],xmm7[2],xmm0[2],xmm7[3],xmm0[3],xmm7[4],xmm0[4],xmm7[5],xmm0[5],xmm7[6],xmm0[6],xmm7[7],xmm0[7]
; SSE2-NEXT:    movdqa %xmm7, %xmm2
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm7 = xmm7[4],xmm0[4],xmm7[5],xmm0[5],xmm7[6],xmm0[6],xmm7[7],xmm0[7]
; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [5,6,7,8]
; SSE2-NEXT:    paddd %xmm0, %xmm7
; SSE2-NEXT:    paddd %xmm0, %xmm4
; SSE2-NEXT:    paddd %xmm0, %xmm1
; SSE2-NEXT:    paddd %xmm0, %xmm3
; SSE2-NEXT:    paddd %xmm0, %xmm14
; SSE2-NEXT:    paddd %xmm0, %xmm13
; SSE2-NEXT:    paddd %xmm0, %xmm11
; SSE2-NEXT:    paddd %xmm0, %xmm9
; SSE2-NEXT:    movdqa %xmm9, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [1,2,3,4]
; SSE2-NEXT:    paddd %xmm0, %xmm2
; SSE2-NEXT:    paddd %xmm0, %xmm5
; SSE2-NEXT:    paddd %xmm0, %xmm8
; SSE2-NEXT:    paddd %xmm0, %xmm6
; SSE2-NEXT:    paddd %xmm0, %xmm15
; SSE2-NEXT:    paddd %xmm0, %xmm10
; SSE2-NEXT:    paddd %xmm0, %xmm12
; SSE2-NEXT:    movdqa -{{[0-9]+}}(%rsp), %xmm9 # 16-byte Reload
; SSE2-NEXT:    paddd %xmm0, %xmm9
; SSE2-NEXT:    movdqa %xmm9, -{{[0-9]+}}(%rsp) # 16-byte Spill
; SSE2-NEXT:    psrld $1, %xmm2
; SSE2-NEXT:    psrld $1, %xmm7
; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
; SSE2-NEXT:    pand %xmm0, %xmm7
; SSE2-NEXT:    pand %xmm0, %xmm2
; SSE2-NEXT:    packuswb %xmm7, %xmm2
; SSE2-NEXT:    psrld $1, %xmm5
; SSE2-NEXT:    psrld $1, %xmm4
; SSE2-NEXT:    pand %xmm0, %xmm4
; SSE2-NEXT:    pand %xmm0, %xmm5
; SSE2-NEXT:    packuswb %xmm4, %xmm5
; SSE2-NEXT:    packuswb %xmm5, %xmm2
; SSE2-NEXT:    psrld $1, %xmm8
; SSE2-NEXT:    psrld $1, %xmm1
; SSE2-NEXT:    pand %xmm0, %xmm1
; SSE2-NEXT:    pand %xmm0, %xmm8
; SSE2-NEXT:    packuswb %xmm1, %xmm8
; SSE2-NEXT:    psrld $1, %xmm6
; SSE2-NEXT:    psrld $1, %xmm3
; SSE2-NEXT:    pand %xmm0, %xmm3
; SSE2-NEXT:    pand %xmm0, %xmm6
; SSE2-NEXT:    packuswb %xmm3, %xmm6
; SSE2-NEXT:    packuswb %xmm6, %xmm8
; SSE2-NEXT:    psrld $1, %xmm15
; SSE2-NEXT:    psrld $1, %xmm14
; SSE2-NEXT:    pand %xmm0, %xmm14
; SSE2-NEXT:    pand %xmm0, %xmm15
; SSE2-NEXT:    packuswb %xmm14, %xmm15
; SSE2-NEXT:    psrld $1, %xmm10
; SSE2-NEXT:    psrld $1, %xmm13
; SSE2-NEXT:    pand %xmm0, %xmm13
; SSE2-NEXT:    pand %xmm0, %xmm10
; SSE2-NEXT:    packuswb %xmm13, %xmm10
; SSE2-NEXT:    packuswb %xmm10, %xmm15
; SSE2-NEXT:    psrld $1, %xmm12
; SSE2-NEXT:    psrld $1, %xmm11
; SSE2-NEXT:    pand %xmm0, %xmm11
; SSE2-NEXT:    pand %xmm0, %xmm12
; SSE2-NEXT:    packuswb %xmm11, %xmm12
; SSE2-NEXT:    movdqa -{{[0-9]+}}(%rsp), %xmm1 # 16-byte Reload
; SSE2-NEXT:    psrld $1, %xmm1
; SSE2-NEXT:    movdqa -{{[0-9]+}}(%rsp), %xmm3 # 16-byte Reload
; SSE2-NEXT:    psrld $1, %xmm3
; SSE2-NEXT:    pand %xmm0, %xmm3
; SSE2-NEXT:    pand %xmm0, %xmm1
; SSE2-NEXT:    packuswb %xmm3, %xmm1
; SSE2-NEXT:    packuswb %xmm1, %xmm12
; SSE2-NEXT:    movdqu %xmm12, (%rax)
; SSE2-NEXT:    movdqu %xmm15, (%rax)
; SSE2-NEXT:    movdqu %xmm8, (%rax)
; SSE2-NEXT:    movdqu %xmm2, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v64i8_const:
; AVX2:       # BB#0:
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm5 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm6 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; AVX2-NEXT:    vmovdqa {{.*#+}} ymm8 = [1,2,3,4,5,6,7,8]
; AVX2-NEXT:    vpaddd %ymm8, %ymm7, %ymm7
; AVX2-NEXT:    vpaddd %ymm8, %ymm6, %ymm6
; AVX2-NEXT:    vpaddd %ymm8, %ymm5, %ymm5
; AVX2-NEXT:    vpaddd %ymm8, %ymm4, %ymm4
; AVX2-NEXT:    vpaddd %ymm8, %ymm3, %ymm3
; AVX2-NEXT:    vpaddd %ymm8, %ymm2, %ymm2
; AVX2-NEXT:    vpaddd %ymm8, %ymm1, %ymm1
; AVX2-NEXT:    vpaddd %ymm8, %ymm0, %ymm0
; AVX2-NEXT:    vpsrld $1, %ymm0, %ymm0
; AVX2-NEXT:    vpsrld $1, %ymm1, %ymm1
; AVX2-NEXT:    vpsrld $1, %ymm2, %ymm8
; AVX2-NEXT:    vpsrld $1, %ymm3, %ymm9
; AVX2-NEXT:    vpsrld $1, %ymm4, %ymm4
; AVX2-NEXT:    vpsrld $1, %ymm5, %ymm5
; AVX2-NEXT:    vpsrld $1, %ymm6, %ymm6
; AVX2-NEXT:    vpsrld $1, %ymm7, %ymm3
; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,1,4,5,8,9,12,13,128,128,128,128,128,128,128,128,0,1,4,5,8,9,12,13,128,128,128,128,128,128,128,128]
; AVX2-NEXT:    vpshufb %ymm2, %ymm3, %ymm3
; AVX2-NEXT:    vpermq {{.*#+}} ymm7 = ymm3[0,2,2,3]
; AVX2-NEXT:    vmovdqa {{.*#+}} xmm3 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
; AVX2-NEXT:    vpshufb %xmm3, %xmm7, %xmm7
; AVX2-NEXT:    vpshufb %ymm2, %ymm6, %ymm6
; AVX2-NEXT:    vpermq {{.*#+}} ymm6 = ymm6[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm6, %xmm6
; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm6 = xmm6[0],xmm7[0]
; AVX2-NEXT:    vpshufb %ymm2, %ymm5, %ymm5
; AVX2-NEXT:    vpermq {{.*#+}} ymm5 = ymm5[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm5, %xmm5
; AVX2-NEXT:    vpshufb %ymm2, %ymm4, %ymm4
; AVX2-NEXT:    vpermq {{.*#+}} ymm4 = ymm4[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm4, %xmm4
; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm4 = xmm4[0],xmm5[0]
; AVX2-NEXT:    vinserti128 $1, %xmm6, %ymm4, %ymm4
; AVX2-NEXT:    vpshufb %ymm2, %ymm9, %ymm5
; AVX2-NEXT:    vpermq {{.*#+}} ymm5 = ymm5[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm5, %xmm5
; AVX2-NEXT:    vpshufb %ymm2, %ymm8, %ymm6
; AVX2-NEXT:    vpermq {{.*#+}} ymm6 = ymm6[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm6, %xmm6
; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm5 = xmm6[0],xmm5[0]
; AVX2-NEXT:    vpshufb %ymm2, %ymm1, %ymm1
; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
; AVX2-NEXT:    vpshufb %ymm2, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
; AVX2-NEXT:    vpshufb %xmm3, %xmm0, %xmm0
; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
; AVX2-NEXT:    vinserti128 $1, %xmm5, %ymm0, %ymm0
; AVX2-NEXT:    vmovdqu %ymm0, (%rax)
; AVX2-NEXT:    vmovdqu %ymm4, (%rax)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v64i8_const:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vpmovzxbd {{.*#+}} zmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm4 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vpaddd %zmm4, %zmm3, %zmm3
; AVX512F-NEXT:    vpaddd %zmm4, %zmm2, %zmm2
; AVX512F-NEXT:    vpaddd %zmm4, %zmm1, %zmm1
; AVX512F-NEXT:    vpaddd %zmm4, %zmm0, %zmm0
; AVX512F-NEXT:    vpsrld $1, %zmm0, %zmm0
; AVX512F-NEXT:    vpsrld $1, %zmm1, %zmm1
; AVX512F-NEXT:    vpsrld $1, %zmm2, %zmm2
; AVX512F-NEXT:    vpsrld $1, %zmm3, %zmm3
; AVX512F-NEXT:    vpmovdb %zmm3, %xmm3
; AVX512F-NEXT:    vpmovdb %zmm2, %xmm2
; AVX512F-NEXT:    vinserti128 $1, %xmm2, %ymm3, %ymm2
; AVX512F-NEXT:    vpmovdb %zmm1, %xmm1
; AVX512F-NEXT:    vpmovdb %zmm0, %xmm0
; AVX512F-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
; AVX512F-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512F-NEXT:    vmovdqu %ymm2, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v64i8_const:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqu8 (%rdi), %zmm0
; AVX512BW-NEXT:    vpavgb {{.*}}(%rip), %zmm0, %zmm0
; AVX512BW-NEXT:    vmovdqu8 %zmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <64 x i8>, <64 x i8>* %a
  %2 = zext <64 x i8> %1 to <64 x i32>
  %3 = add nuw nsw <64 x i32> %2, <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
  %4 = lshr <64 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %5 = trunc <64 x i32> %4 to <64 x i8>
  store <64 x i8> %5, <64 x i8>* undef, align 4
  ret void
}

define void @avg_v4i16_const(<4 x i16>* %a) {
; SSE2-LABEL: avg_v4i16_const:
; SSE2:       # BB#0:
; SSE2-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
; SSE2-NEXT:    pavgw {{.*}}(%rip), %xmm0
; SSE2-NEXT:    movq %xmm0, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v4i16_const:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX2-NEXT:    vpavgw {{.*}}(%rip), %xmm0, %xmm0
; AVX2-NEXT:    vmovq %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v4i16_const:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512F-NEXT:    vpavgw {{.*}}(%rip), %xmm0, %xmm0
; AVX512F-NEXT:    vmovq %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v4i16_const:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
; AVX512BW-NEXT:    vpavgw {{.*}}(%rip), %xmm0, %xmm0
; AVX512BW-NEXT:    vmovq %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <4 x i16>, <4 x i16>* %a
  %2 = zext <4 x i16> %1 to <4 x i32>
  %3 = add nuw nsw <4 x i32> %2, <i32 1, i32 2, i32 3, i32 4>
  %4 = lshr <4 x i32> %3, <i32 1, i32 1, i32 1, i32 1>
  %5 = trunc <4 x i32> %4 to <4 x i16>
  store <4 x i16> %5, <4 x i16>* undef, align 4
  ret void
}

define void @avg_v8i16_const(<8 x i16>* %a) {
; SSE2-LABEL: avg_v8i16_const:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm0
; SSE2-NEXT:    pavgw {{.*}}(%rip), %xmm0
; SSE2-NEXT:    movdqu %xmm0, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v8i16_const:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovdqa (%rdi), %xmm0
; AVX2-NEXT:    vpavgw {{.*}}(%rip), %xmm0, %xmm0
; AVX2-NEXT:    vmovdqu %xmm0, (%rax)
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v8i16_const:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512F-NEXT:    vpavgw {{.*}}(%rip), %xmm0, %xmm0
; AVX512F-NEXT:    vmovdqu %xmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v8i16_const:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %xmm0
; AVX512BW-NEXT:    vpavgw {{.*}}(%rip), %xmm0, %xmm0
; AVX512BW-NEXT:    vmovdqu %xmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <8 x i16>, <8 x i16>* %a
  %2 = zext <8 x i16> %1 to <8 x i32>
  %3 = add nuw nsw <8 x i32> %2, <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
  %4 = lshr <8 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %5 = trunc <8 x i32> %4 to <8 x i16>
  store <8 x i16> %5, <8 x i16>* undef, align 4
  ret void
}

define void @avg_v16i16_const(<16 x i16>* %a) {
; SSE2-LABEL: avg_v16i16_const:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm3
; SSE2-NEXT:    movdqa 16(%rdi), %xmm0
; SSE2-NEXT:    pxor %xmm4, %xmm4
; SSE2-NEXT:    movdqa %xmm0, %xmm1
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1],xmm1[2],xmm4[2],xmm1[3],xmm4[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
; SSE2-NEXT:    movdqa %xmm3, %xmm2
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1],xmm2[2],xmm4[2],xmm2[3],xmm4[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm4[4],xmm3[5],xmm4[5],xmm3[6],xmm4[6],xmm3[7],xmm4[7]
; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [5,6,7,8]
; SSE2-NEXT:    paddd %xmm4, %xmm3
; SSE2-NEXT:    movdqa {{.*#+}} xmm5 = [1,2,3,4]
; SSE2-NEXT:    paddd %xmm5, %xmm2
; SSE2-NEXT:    paddd %xmm4, %xmm0
; SSE2-NEXT:    paddd %xmm5, %xmm1
; SSE2-NEXT:    psrld $1, %xmm1
; SSE2-NEXT:    psrld $1, %xmm0
; SSE2-NEXT:    psrld $1, %xmm2
; SSE2-NEXT:    psrld $1, %xmm3
; SSE2-NEXT:    pslld $16, %xmm3
; SSE2-NEXT:    psrad $16, %xmm3
; SSE2-NEXT:    pslld $16, %xmm2
; SSE2-NEXT:    psrad $16, %xmm2
; SSE2-NEXT:    packssdw %xmm3, %xmm2
; SSE2-NEXT:    pslld $16, %xmm0
; SSE2-NEXT:    psrad $16, %xmm0
; SSE2-NEXT:    pslld $16, %xmm1
; SSE2-NEXT:    psrad $16, %xmm1
; SSE2-NEXT:    packssdw %xmm0, %xmm1
; SSE2-NEXT:    movdqu %xmm1, (%rax)
; SSE2-NEXT:    movdqu %xmm2, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v16i16_const:
; AVX2:       # BB#0:
; AVX2-NEXT:    vmovdqa (%rdi), %ymm0
; AVX2-NEXT:    vpavgw {{.*}}(%rip), %ymm0, %ymm0
; AVX2-NEXT:    vmovdqu %ymm0, (%rax)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v16i16_const:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vmovdqa (%rdi), %ymm0
; AVX512F-NEXT:    vpavgw {{.*}}(%rip), %ymm0, %ymm0
; AVX512F-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v16i16_const:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqa (%rdi), %ymm0
; AVX512BW-NEXT:    vpavgw {{.*}}(%rip), %ymm0, %ymm0
; AVX512BW-NEXT:    vmovdqu %ymm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <16 x i16>, <16 x i16>* %a
  %2 = zext <16 x i16> %1 to <16 x i32>
  %3 = add nuw nsw <16 x i32> %2, <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
  %4 = lshr <16 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %5 = trunc <16 x i32> %4 to <16 x i16>
  store <16 x i16> %5, <16 x i16>* undef, align 4
  ret void
}

define void @avg_v32i16_const(<32 x i16>* %a) {
; SSE2-LABEL: avg_v32i16_const:
; SSE2:       # BB#0:
; SSE2-NEXT:    movdqa (%rdi), %xmm7
; SSE2-NEXT:    movdqa 16(%rdi), %xmm6
; SSE2-NEXT:    movdqa 32(%rdi), %xmm4
; SSE2-NEXT:    movdqa 48(%rdi), %xmm0
; SSE2-NEXT:    pxor %xmm8, %xmm8
; SSE2-NEXT:    movdqa %xmm0, %xmm1
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm8[0],xmm1[1],xmm8[1],xmm1[2],xmm8[2],xmm1[3],xmm8[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm8[4],xmm0[5],xmm8[5],xmm0[6],xmm8[6],xmm0[7],xmm8[7]
; SSE2-NEXT:    movdqa %xmm4, %xmm2
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm8[0],xmm2[1],xmm8[1],xmm2[2],xmm8[2],xmm2[3],xmm8[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm8[4],xmm4[5],xmm8[5],xmm4[6],xmm8[6],xmm4[7],xmm8[7]
; SSE2-NEXT:    movdqa %xmm6, %xmm3
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm8[0],xmm3[1],xmm8[1],xmm3[2],xmm8[2],xmm3[3],xmm8[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm8[4],xmm6[5],xmm8[5],xmm6[6],xmm8[6],xmm6[7],xmm8[7]
; SSE2-NEXT:    movdqa %xmm7, %xmm5
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm8[0],xmm5[1],xmm8[1],xmm5[2],xmm8[2],xmm5[3],xmm8[3]
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm7 = xmm7[4],xmm8[4],xmm7[5],xmm8[5],xmm7[6],xmm8[6],xmm7[7],xmm8[7]
; SSE2-NEXT:    movdqa {{.*#+}} xmm8 = [5,6,7,8]
; SSE2-NEXT:    paddd %xmm8, %xmm7
; SSE2-NEXT:    movdqa {{.*#+}} xmm9 = [1,2,3,4]
; SSE2-NEXT:    paddd %xmm9, %xmm5
; SSE2-NEXT:    paddd %xmm8, %xmm6
; SSE2-NEXT:    paddd %xmm9, %xmm3
; SSE2-NEXT:    paddd %xmm8, %xmm4
; SSE2-NEXT:    paddd %xmm9, %xmm2
; SSE2-NEXT:    paddd %xmm8, %xmm0
; SSE2-NEXT:    paddd %xmm9, %xmm1
; SSE2-NEXT:    psrld $1, %xmm1
; SSE2-NEXT:    psrld $1, %xmm0
; SSE2-NEXT:    psrld $1, %xmm2
; SSE2-NEXT:    psrld $1, %xmm4
; SSE2-NEXT:    psrld $1, %xmm3
; SSE2-NEXT:    psrld $1, %xmm6
; SSE2-NEXT:    psrld $1, %xmm5
; SSE2-NEXT:    psrld $1, %xmm7
; SSE2-NEXT:    pslld $16, %xmm7
; SSE2-NEXT:    psrad $16, %xmm7
; SSE2-NEXT:    pslld $16, %xmm5
; SSE2-NEXT:    psrad $16, %xmm5
; SSE2-NEXT:    packssdw %xmm7, %xmm5
; SSE2-NEXT:    pslld $16, %xmm6
; SSE2-NEXT:    psrad $16, %xmm6
; SSE2-NEXT:    pslld $16, %xmm3
; SSE2-NEXT:    psrad $16, %xmm3
; SSE2-NEXT:    packssdw %xmm6, %xmm3
; SSE2-NEXT:    pslld $16, %xmm4
; SSE2-NEXT:    psrad $16, %xmm4
; SSE2-NEXT:    pslld $16, %xmm2
; SSE2-NEXT:    psrad $16, %xmm2
; SSE2-NEXT:    packssdw %xmm4, %xmm2
; SSE2-NEXT:    pslld $16, %xmm0
; SSE2-NEXT:    psrad $16, %xmm0
; SSE2-NEXT:    pslld $16, %xmm1
; SSE2-NEXT:    psrad $16, %xmm1
; SSE2-NEXT:    packssdw %xmm0, %xmm1
; SSE2-NEXT:    movdqu %xmm1, (%rax)
; SSE2-NEXT:    movdqu %xmm2, (%rax)
; SSE2-NEXT:    movdqu %xmm3, (%rax)
; SSE2-NEXT:    movdqu %xmm5, (%rax)
; SSE2-NEXT:    retq
;
; AVX2-LABEL: avg_v32i16_const:
; AVX2:       # BB#0:
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm2 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = [1,2,3,4,5,6,7,8]
; AVX2-NEXT:    vpaddd %ymm4, %ymm3, %ymm3
; AVX2-NEXT:    vpaddd %ymm4, %ymm2, %ymm2
; AVX2-NEXT:    vpaddd %ymm4, %ymm1, %ymm1
; AVX2-NEXT:    vpaddd %ymm4, %ymm0, %ymm0
; AVX2-NEXT:    vpsrld $1, %ymm0, %ymm0
; AVX2-NEXT:    vpsrld $1, %ymm1, %ymm1
; AVX2-NEXT:    vpsrld $1, %ymm2, %ymm2
; AVX2-NEXT:    vpsrld $1, %ymm3, %ymm3
; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = [0,1,4,5,8,9,12,13,128,128,128,128,128,128,128,128,0,1,4,5,8,9,12,13,128,128,128,128,128,128,128,128]
; AVX2-NEXT:    vpshufb %ymm4, %ymm3, %ymm3
; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
; AVX2-NEXT:    vpshufb %ymm4, %ymm2, %ymm2
; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
; AVX2-NEXT:    vinserti128 $1, %xmm2, %ymm3, %ymm2
; AVX2-NEXT:    vpshufb %ymm4, %ymm1, %ymm1
; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
; AVX2-NEXT:    vpshufb %ymm4, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
; AVX2-NEXT:    vmovdqu %ymm0, (%rax)
; AVX2-NEXT:    vmovdqu %ymm2, (%rax)
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: avg_v32i16_const:
; AVX512F:       # BB#0:
; AVX512F-NEXT:    vpmovzxwd {{.*#+}} zmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero
; AVX512F-NEXT:    vpmovzxwd {{.*#+}} zmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero
; AVX512F-NEXT:    vbroadcasti64x4 {{.*#+}} zmm2 = mem[0,1,2,3,0,1,2,3]
; AVX512F-NEXT:    vpaddd %zmm2, %zmm1, %zmm1
; AVX512F-NEXT:    vpaddd %zmm2, %zmm0, %zmm0
; AVX512F-NEXT:    vpsrld $1, %zmm0, %zmm0
; AVX512F-NEXT:    vpsrld $1, %zmm1, %zmm1
; AVX512F-NEXT:    vpmovdw %zmm1, (%rax)
; AVX512F-NEXT:    vpmovdw %zmm0, (%rax)
; AVX512F-NEXT:    retq
;
; AVX512BW-LABEL: avg_v32i16_const:
; AVX512BW:       # BB#0:
; AVX512BW-NEXT:    vmovdqu16 (%rdi), %zmm0
; AVX512BW-NEXT:    vpavgw {{.*}}(%rip), %zmm0, %zmm0
; AVX512BW-NEXT:    vmovdqu16 %zmm0, (%rax)
; AVX512BW-NEXT:    retq
  %1 = load <32 x i16>, <32 x i16>* %a
  %2 = zext <32 x i16> %1 to <32 x i32>
  %3 = add nuw nsw <32 x i32> %2, <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
  %4 = lshr <32 x i32> %3, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %5 = trunc <32 x i32> %4 to <32 x i16>
  store <32 x i16> %5, <32 x i16>* undef, align 4
  ret void
}