[X86][AVX] Extended vector average tests

Added AVX1 tests and merged AVX1/AVX2/AVX512 checks where possible


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@306107 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 917 insertions, 411 deletions

diff --git a/test/CodeGen/X86/avg.ll b/test/CodeGen/X86/avg.ll
index 634007167f0..d4712becd09 100644
--- a/test/CodeGen/X86/avg.ll
+++ b/test/CodeGen/X86/avg.ll
@@ -1,8 +1,9 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefix=SSE2
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=AVX2
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefix=AVX512F
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw | FileCheck %s --check-prefix=AVX512BW
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=AVX --check-prefix=AVX1
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=AVX --check-prefix=AVX2
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefix=AVX --check-prefix=AVX512 --check-prefix=AVX512F
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw | FileCheck %s --check-prefix=AVX --check-prefix=AVX512 --check-prefix=AVX512BW
 
 define void @avg_v4i8(<4 x i8>* %a, <4 x i8>* %b) {
 ; SSE2-LABEL: avg_v4i8:
@@ -13,29 +14,13 @@ define void @avg_v4i8(<4 x i8>* %a, <4 x i8>* %b) {
 ; 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
+; AVX-LABEL: avg_v4i8:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX-NEXT:    vpavgb %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vmovd %xmm0, (%rax)
+; AVX-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>
@@ -57,29 +42,13 @@ define void @avg_v8i8(<8 x i8>* %a, <8 x i8>* %b) {
 ; 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
+; AVX-LABEL: avg_v8i8:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX-NEXT:    vpavgb %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vmovq %xmm0, (%rax)
+; AVX-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>
@@ -100,26 +69,12 @@ define void @avg_v16i8(<16 x i8>* %a, <16 x i8>* %b) {
 ; 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
+; AVX-LABEL: avg_v16i8:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovdqa (%rsi), %xmm0
+; AVX-NEXT:    vpavgb (%rdi), %xmm0, %xmm0
+; AVX-NEXT:    vmovdqu %xmm0, (%rax)
+; AVX-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>
@@ -221,6 +176,69 @@ define void @avg_v32i8(<32 x i8>* %a, <32 x i8>* %b) {
 ; SSE2-NEXT:    movdqu %xmm4, (%rax)
 ; SSE2-NEXT:    retq
 ;
+; AVX1-LABEL: avg_v32i8:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm8 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm6 = [1,1,1,1]
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm1, %xmm1
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm2, %xmm9
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm4, %xmm4
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm5, %xmm5
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm8, %xmm7
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm2, %xmm2
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm6 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm2, %xmm6, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm0, %xmm0
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm6 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
+; AVX1-NEXT:    vpand %xmm6, %xmm0, %xmm0
+; AVX1-NEXT:    vpand %xmm6, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm9, %xmm3
+; AVX1-NEXT:    vpand %xmm6, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm6, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm5, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm4, %xmm3
+; AVX1-NEXT:    vpand %xmm6, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm6, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm7, %xmm3
+; AVX1-NEXT:    vpand %xmm6, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm6, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vmovups %ymm0, (%rax)
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
 ; AVX2-LABEL: avg_v32i8:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vmovdqa (%rsi), %ymm0
@@ -229,21 +247,13 @@ define void @avg_v32i8(<32 x i8>* %a, <32 x i8>* %b) {
 ; 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:    vzeroupper
-; 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:    vzeroupper
-; AVX512BW-NEXT:    retq
+; AVX512-LABEL: avg_v32i8:
+; AVX512:       # BB#0:
+; AVX512-NEXT:    vmovdqa (%rsi), %ymm0
+; AVX512-NEXT:    vpavgb (%rdi), %ymm0, %ymm0
+; AVX512-NEXT:    vmovdqu %ymm0, (%rax)
+; AVX512-NEXT:    vzeroupper
+; AVX512-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>
@@ -454,6 +464,131 @@ define void @avg_v64i8(<64 x i8>* %a, <64 x i8>* %b) {
 ; SSE2-NEXT:    addq $152, %rsp
 ; SSE2-NEXT:    retq
 ;
+; AVX1-LABEL: avg_v64i8:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm10 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm11 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm15 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm6 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm8 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm12 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm13 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vmovdqa %xmm0, -{{[0-9]+}}(%rsp) # 16-byte Spill
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vmovdqa %xmm0, -{{[0-9]+}}(%rsp) # 16-byte Spill
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm0 = [1,1,1,1]
+; AVX1-NEXT:    vpaddd %xmm0, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm5, %xmm14
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm7, %xmm7
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm10, %xmm4
+; AVX1-NEXT:    vmovdqa %xmm4, -{{[0-9]+}}(%rsp) # 16-byte Spill
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm2, %xmm9
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm11, %xmm11
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm15, %xmm15
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm6, %xmm6
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm8, %xmm2
+; AVX1-NEXT:    vmovdqa %xmm2, -{{[0-9]+}}(%rsp) # 16-byte Spill
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm1, %xmm8
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm1, %xmm1
+; AVX1-NEXT:    vpaddd %xmm1, %xmm12, %xmm12
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm1, %xmm1
+; AVX1-NEXT:    vpaddd %xmm1, %xmm13, %xmm4
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm1, %xmm1
+; AVX1-NEXT:    vpaddd -{{[0-9]+}}(%rsp), %xmm1, %xmm13 # 16-byte Folded Reload
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm1, %xmm1
+; AVX1-NEXT:    vpaddd -{{[0-9]+}}(%rsp), %xmm1, %xmm5 # 16-byte Folded Reload
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm1, %xmm1
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm1, %xmm2, %xmm10
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm0, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm7, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm14, %xmm14
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm7 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
+; AVX1-NEXT:    vpand %xmm7, %xmm14, %xmm14
+; AVX1-NEXT:    vpand %xmm7, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm14, %xmm0, %xmm14
+; AVX1-NEXT:    vpsrld $1, %xmm9, %xmm0
+; AVX1-NEXT:    vmovdqa -{{[0-9]+}}(%rsp), %xmm2 # 16-byte Reload
+; AVX1-NEXT:    vpsrld $1, %xmm2, %xmm2
+; AVX1-NEXT:    vpand %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpand %xmm7, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm14, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm11, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm6, %xmm3
+; AVX1-NEXT:    vpsrld $1, %xmm15, %xmm6
+; AVX1-NEXT:    vpand %xmm7, %xmm6, %xmm6
+; AVX1-NEXT:    vpand %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm6, %xmm3, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm2, %ymm0
+; AVX1-NEXT:    vpsrld $1, %xmm8, %xmm2
+; AVX1-NEXT:    vmovdqa -{{[0-9]+}}(%rsp), %xmm3 # 16-byte Reload
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm4, %xmm3
+; AVX1-NEXT:    vpsrld $1, %xmm12, %xmm4
+; AVX1-NEXT:    vpand %xmm7, %xmm4, %xmm4
+; AVX1-NEXT:    vpand %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm5, %xmm3
+; AVX1-NEXT:    vpsrld $1, %xmm13, %xmm4
+; AVX1-NEXT:    vpand %xmm7, %xmm4, %xmm4
+; AVX1-NEXT:    vpand %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpsrld $1, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm10, %xmm4
+; AVX1-NEXT:    vpand %xmm7, %xmm4, %xmm4
+; AVX1-NEXT:    vpand %xmm7, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm1
+; AVX1-NEXT:    vmovups %ymm1, (%rax)
+; AVX1-NEXT:    vmovups %ymm0, (%rax)
+; AVX1-NEXT:    vzeroupper
+; AVX1-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
@@ -596,29 +731,13 @@ define void @avg_v4i16(<4 x i16>* %a, <4 x i16>* %b) {
 ; 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
+; AVX-LABEL: avg_v4i16:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX-NEXT:    vpavgw %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vmovq %xmm0, (%rax)
+; AVX-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>
@@ -639,26 +758,12 @@ define void @avg_v8i16(<8 x i16>* %a, <8 x i16>* %b) {
 ; 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
+; AVX-LABEL: avg_v8i16:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovdqa (%rsi), %xmm0
+; AVX-NEXT:    vpavgw (%rdi), %xmm0, %xmm0
+; AVX-NEXT:    vmovdqu %xmm0, (%rax)
+; AVX-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>
@@ -718,6 +823,41 @@ define void @avg_v16i16(<16 x i16>* %a, <16 x i16>* %b) {
 ; SSE2-NEXT:    movdqu %xmm0, (%rax)
 ; SSE2-NEXT:    retq
 ;
+; AVX1-LABEL: avg_v16i16:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm8 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm4 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm5 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm6 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [1,1,1,1]
+; AVX1-NEXT:    vpaddd %xmm3, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpaddd %xmm3, %xmm5, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpaddd %xmm3, %xmm6, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddd %xmm3, %xmm7, %xmm3
+; AVX1-NEXT:    vpaddd %xmm3, %xmm8, %xmm3
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm3
+; AVX1-NEXT:    vpsrld $1, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],xmm4[1],xmm0[2],xmm4[3],xmm0[4],xmm4[5],xmm0[6],xmm4[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0],xmm4[1],xmm1[2],xmm4[3],xmm1[4],xmm4[5],xmm1[6],xmm4[7]
+; AVX1-NEXT:    vpackusdw %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0],xmm4[1],xmm2[2],xmm4[3],xmm2[4],xmm4[5],xmm2[6],xmm4[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm3[0],xmm4[1],xmm3[2],xmm4[3],xmm3[4],xmm4[5],xmm3[6],xmm4[7]
+; AVX1-NEXT:    vpackusdw %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vmovups %ymm0, (%rax)
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
 ; AVX2-LABEL: avg_v16i16:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vmovdqa (%rsi), %ymm0
@@ -726,21 +866,13 @@ define void @avg_v16i16(<16 x i16>* %a, <16 x i16>* %b) {
 ; 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:    vzeroupper
-; 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:    vzeroupper
-; AVX512BW-NEXT:    retq
+; AVX512-LABEL: avg_v16i16:
+; AVX512:       # BB#0:
+; AVX512-NEXT:    vmovdqa (%rsi), %ymm0
+; AVX512-NEXT:    vpavgw (%rdi), %ymm0, %ymm0
+; AVX512-NEXT:    vmovdqu %ymm0, (%rax)
+; AVX512-NEXT:    vzeroupper
+; AVX512-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>
@@ -841,6 +973,69 @@ define void @avg_v32i16(<32 x i16>* %a, <32 x i16>* %b) {
 ; SSE2-NEXT:    movdqu %xmm8, (%rax)
 ; SSE2-NEXT:    retq
 ;
+; AVX1-LABEL: avg_v32i16:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm4 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm5 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm8 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm6 = [1,1,1,1]
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm1, %xmm1
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm4, %xmm9
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm5, %xmm5
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm8, %xmm7
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm4 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm4, %xmm4
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm6 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm4, %xmm6, %xmm4
+; AVX1-NEXT:    vpsrld $1, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm6, %xmm6, %xmm6
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],xmm6[1],xmm0[2],xmm6[3],xmm0[4],xmm6[5],xmm0[6],xmm6[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0],xmm6[1],xmm1[2],xmm6[3],xmm1[4],xmm6[5],xmm1[6],xmm6[7]
+; AVX1-NEXT:    vpackusdw %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm2, %xmm2
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm2[0],xmm6[1],xmm2[2],xmm6[3],xmm2[4],xmm6[5],xmm2[6],xmm6[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0],xmm6[1],xmm1[2],xmm6[3],xmm1[4],xmm6[5],xmm1[6],xmm6[7]
+; AVX1-NEXT:    vpackusdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vpsrld $1, %xmm5, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm9, %xmm2
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm2[0],xmm6[1],xmm2[2],xmm6[3],xmm2[4],xmm6[5],xmm2[6],xmm6[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0],xmm6[1],xmm1[2],xmm6[3],xmm1[4],xmm6[5],xmm1[6],xmm6[7]
+; AVX1-NEXT:    vpackusdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm4, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm7, %xmm3
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm3 = xmm3[0],xmm6[1],xmm3[2],xmm6[3],xmm3[4],xmm6[5],xmm3[6],xmm6[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm2[0],xmm6[1],xmm2[2],xmm6[3],xmm2[4],xmm6[5],xmm2[6],xmm6[7]
+; AVX1-NEXT:    vpackusdw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm1
+; AVX1-NEXT:    vmovups %ymm1, (%rax)
+; AVX1-NEXT:    vmovups %ymm0, (%rax)
+; AVX1-NEXT:    vzeroupper
+; AVX1-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
@@ -926,29 +1121,13 @@ define void @avg_v4i8_2(<4 x i8>* %a, <4 x i8>* %b) {
 ; 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
+; AVX-LABEL: avg_v4i8_2:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX-NEXT:    vpavgb %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vmovd %xmm0, (%rax)
+; AVX-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>
@@ -970,29 +1149,13 @@ define void @avg_v8i8_2(<8 x i8>* %a, <8 x i8>* %b) {
 ; 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
+; AVX-LABEL: avg_v8i8_2:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX-NEXT:    vpavgb %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vmovq %xmm0, (%rax)
+; AVX-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>
@@ -1013,26 +1176,12 @@ define void @avg_v16i8_2(<16 x i8>* %a, <16 x i8>* %b) {
 ; 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
+; AVX-LABEL: avg_v16i8_2:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX-NEXT:    vpavgb (%rsi), %xmm0, %xmm0
+; AVX-NEXT:    vmovdqu %xmm0, (%rax)
+; AVX-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>
@@ -1134,6 +1283,69 @@ define void @avg_v32i8_2(<32 x i8>* %a, <32 x i8>* %b) {
 ; SSE2-NEXT:    movdqu %xmm4, (%rax)
 ; SSE2-NEXT:    retq
 ;
+; AVX1-LABEL: avg_v32i8_2:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm8 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm6 = [1,1,1,1]
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm1, %xmm1
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm2, %xmm9
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm4, %xmm4
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm5, %xmm5
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm8, %xmm7
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm2, %xmm2
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm6 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm2, %xmm6, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm0, %xmm0
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm6 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
+; AVX1-NEXT:    vpand %xmm6, %xmm0, %xmm0
+; AVX1-NEXT:    vpand %xmm6, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm9, %xmm3
+; AVX1-NEXT:    vpand %xmm6, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm6, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm5, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm4, %xmm3
+; AVX1-NEXT:    vpand %xmm6, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm6, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm7, %xmm3
+; AVX1-NEXT:    vpand %xmm6, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm6, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vmovups %ymm0, (%rax)
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
 ; AVX2-LABEL: avg_v32i8_2:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vmovdqa (%rdi), %ymm0
@@ -1142,21 +1354,13 @@ define void @avg_v32i8_2(<32 x i8>* %a, <32 x i8>* %b) {
 ; 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:    vzeroupper
-; 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:    vzeroupper
-; AVX512BW-NEXT:    retq
+; AVX512-LABEL: avg_v32i8_2:
+; AVX512:       # BB#0:
+; AVX512-NEXT:    vmovdqa (%rdi), %ymm0
+; AVX512-NEXT:    vpavgb (%rsi), %ymm0, %ymm0
+; AVX512-NEXT:    vmovdqu %ymm0, (%rax)
+; AVX512-NEXT:    vzeroupper
+; AVX512-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>
@@ -1325,6 +1529,117 @@ define void @avg_v64i8_2(<64 x i8>* %a, <64 x i8>* %b) {
 ; SSE2-NEXT:    movdqu %xmm10, (%rax)
 ; SSE2-NEXT:    retq
 ;
+; AVX1-LABEL: avg_v64i8_2:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm14 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm8 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm11 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm12 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm15 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm13 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm9 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm10 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm6 = [1,1,1,1]
+; AVX1-NEXT:    vpaddd %xmm6, %xmm0, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm0, %xmm0
+; AVX1-NEXT:    vmovdqa %xmm0, -{{[0-9]+}}(%rsp) # 16-byte Spill
+; AVX1-NEXT:    vpaddd %xmm6, %xmm1, %xmm0
+; AVX1-NEXT:    vpaddd %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpaddd %xmm6, %xmm2, %xmm1
+; AVX1-NEXT:    vpaddd %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vmovdqa %xmm1, -{{[0-9]+}}(%rsp) # 16-byte Spill
+; AVX1-NEXT:    vpaddd %xmm6, %xmm3, %xmm1
+; AVX1-NEXT:    vpaddd %xmm1, %xmm3, %xmm1
+; AVX1-NEXT:    vpaddd %xmm6, %xmm4, %xmm2
+; AVX1-NEXT:    vpaddd %xmm2, %xmm4, %xmm2
+; AVX1-NEXT:    vmovdqa %xmm2, -{{[0-9]+}}(%rsp) # 16-byte Spill
+; AVX1-NEXT:    vpaddd %xmm6, %xmm5, %xmm2
+; AVX1-NEXT:    vpaddd %xmm2, %xmm5, %xmm2
+; AVX1-NEXT:    vpaddd %xmm6, %xmm14, %xmm3
+; AVX1-NEXT:    vpaddd %xmm3, %xmm14, %xmm14
+; AVX1-NEXT:    vpaddd %xmm6, %xmm8, %xmm3
+; AVX1-NEXT:    vpaddd %xmm3, %xmm8, %xmm5
+; AVX1-NEXT:    vpaddd %xmm6, %xmm11, %xmm3
+; AVX1-NEXT:    vpaddd %xmm3, %xmm11, %xmm3
+; AVX1-NEXT:    vmovdqa %xmm3, -{{[0-9]+}}(%rsp) # 16-byte Spill
+; AVX1-NEXT:    vpaddd %xmm6, %xmm12, %xmm3
+; AVX1-NEXT:    vpaddd %xmm3, %xmm12, %xmm8
+; AVX1-NEXT:    vpaddd %xmm6, %xmm15, %xmm3
+; AVX1-NEXT:    vpaddd %xmm3, %xmm15, %xmm11
+; AVX1-NEXT:    vpaddd %xmm6, %xmm13, %xmm3
+; AVX1-NEXT:    vpaddd %xmm3, %xmm13, %xmm13
+; AVX1-NEXT:    vpaddd %xmm6, %xmm9, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm9, %xmm12
+; AVX1-NEXT:    vpaddd %xmm6, %xmm10, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm10, %xmm4
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm3, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm3, %xmm15
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm3, %xmm6
+; AVX1-NEXT:    vpaddd %xmm6, %xmm3, %xmm6
+; AVX1-NEXT:    vpsrld $1, %xmm0, %xmm0
+; AVX1-NEXT:    vmovdqa -{{[0-9]+}}(%rsp), %xmm3 # 16-byte Reload
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm3
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm7 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
+; AVX1-NEXT:    vpand %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm1, %xmm1
+; AVX1-NEXT:    vmovdqa -{{[0-9]+}}(%rsp), %xmm3 # 16-byte Reload
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm2, %xmm1
+; AVX1-NEXT:    vmovdqa -{{[0-9]+}}(%rsp), %xmm2 # 16-byte Reload
+; AVX1-NEXT:    vpsrld $1, %xmm2, %xmm2
+; AVX1-NEXT:    vpand %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpand %xmm7, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm5, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm14, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vpsrld $1, %xmm8, %xmm1
+; AVX1-NEXT:    vmovdqa -{{[0-9]+}}(%rsp), %xmm2 # 16-byte Reload
+; AVX1-NEXT:    vpsrld $1, %xmm2, %xmm2
+; AVX1-NEXT:    vpand %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpand %xmm7, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm13, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm11, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm4, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm12, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm6, %xmm3
+; AVX1-NEXT:    vpsrld $1, %xmm15, %xmm4
+; AVX1-NEXT:    vpand %xmm7, %xmm4, %xmm4
+; AVX1-NEXT:    vpand %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm1
+; AVX1-NEXT:    vmovups %ymm1, (%rax)
+; AVX1-NEXT:    vmovups %ymm0, (%rax)
+; AVX1-NEXT:    vzeroupper
+; AVX1-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
@@ -1456,29 +1771,13 @@ define void @avg_v4i16_2(<4 x i16>* %a, <4 x i16>* %b) {
 ; 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
+; AVX-LABEL: avg_v4i16_2:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX-NEXT:    vpavgw %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vmovq %xmm0, (%rax)
+; AVX-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>
@@ -1499,26 +1798,12 @@ define void @avg_v8i16_2(<8 x i16>* %a, <8 x i16>* %b) {
 ; 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
+; AVX-LABEL: avg_v8i16_2:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX-NEXT:    vpavgw (%rsi), %xmm0, %xmm0
+; AVX-NEXT:    vmovdqu %xmm0, (%rax)
+; AVX-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>
@@ -1578,6 +1863,41 @@ define void @avg_v16i16_2(<16 x i16>* %a, <16 x i16>* %b) {
 ; SSE2-NEXT:    movdqu %xmm0, (%rax)
 ; SSE2-NEXT:    retq
 ;
+; AVX1-LABEL: avg_v16i16_2:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm8 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm4 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm5 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm6 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [1,1,1,1]
+; AVX1-NEXT:    vpaddd %xmm3, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpaddd %xmm3, %xmm5, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpaddd %xmm3, %xmm6, %xmm4
+; AVX1-NEXT:    vpaddd %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddd %xmm3, %xmm7, %xmm3
+; AVX1-NEXT:    vpaddd %xmm3, %xmm8, %xmm3
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm3
+; AVX1-NEXT:    vpsrld $1, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],xmm4[1],xmm0[2],xmm4[3],xmm0[4],xmm4[5],xmm0[6],xmm4[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0],xmm4[1],xmm1[2],xmm4[3],xmm1[4],xmm4[5],xmm1[6],xmm4[7]
+; AVX1-NEXT:    vpackusdw %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0],xmm4[1],xmm2[2],xmm4[3],xmm2[4],xmm4[5],xmm2[6],xmm4[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm3[0],xmm4[1],xmm3[2],xmm4[3],xmm3[4],xmm4[5],xmm3[6],xmm4[7]
+; AVX1-NEXT:    vpackusdw %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vmovups %ymm0, (%rax)
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
 ; AVX2-LABEL: avg_v16i16_2:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vmovdqa (%rdi), %ymm0
@@ -1586,21 +1906,13 @@ define void @avg_v16i16_2(<16 x i16>* %a, <16 x i16>* %b) {
 ; 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:    vzeroupper
-; 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:    vzeroupper
-; AVX512BW-NEXT:    retq
+; AVX512-LABEL: avg_v16i16_2:
+; AVX512:       # BB#0:
+; AVX512-NEXT:    vmovdqa (%rdi), %ymm0
+; AVX512-NEXT:    vpavgw (%rsi), %ymm0, %ymm0
+; AVX512-NEXT:    vmovdqu %ymm0, (%rax)
+; AVX512-NEXT:    vzeroupper
+; AVX512-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>
@@ -1701,6 +2013,69 @@ define void @avg_v32i16_2(<32 x i16>* %a, <32 x i16>* %b) {
 ; SSE2-NEXT:    movdqu %xmm8, (%rax)
 ; SSE2-NEXT:    retq
 ;
+; AVX1-LABEL: avg_v32i16_2:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm4 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm5 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm8 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm6 = [1,1,1,1]
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm1, %xmm1
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm4, %xmm9
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm5, %xmm5
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm7, %xmm7
+; AVX1-NEXT:    vpaddd %xmm7, %xmm8, %xmm7
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm4 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm6, %xmm4, %xmm4
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm6 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpaddd %xmm4, %xmm6, %xmm4
+; AVX1-NEXT:    vpsrld $1, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm6, %xmm6, %xmm6
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],xmm6[1],xmm0[2],xmm6[3],xmm0[4],xmm6[5],xmm0[6],xmm6[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0],xmm6[1],xmm1[2],xmm6[3],xmm1[4],xmm6[5],xmm1[6],xmm6[7]
+; AVX1-NEXT:    vpackusdw %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm2, %xmm2
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm2[0],xmm6[1],xmm2[2],xmm6[3],xmm2[4],xmm6[5],xmm2[6],xmm6[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0],xmm6[1],xmm1[2],xmm6[3],xmm1[4],xmm6[5],xmm1[6],xmm6[7]
+; AVX1-NEXT:    vpackusdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vpsrld $1, %xmm5, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm9, %xmm2
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm2[0],xmm6[1],xmm2[2],xmm6[3],xmm2[4],xmm6[5],xmm2[6],xmm6[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0],xmm6[1],xmm1[2],xmm6[3],xmm1[4],xmm6[5],xmm1[6],xmm6[7]
+; AVX1-NEXT:    vpackusdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm4, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm7, %xmm3
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm3 = xmm3[0],xmm6[1],xmm3[2],xmm6[3],xmm3[4],xmm6[5],xmm3[6],xmm6[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm2[0],xmm6[1],xmm2[2],xmm6[3],xmm2[4],xmm6[5],xmm2[6],xmm6[7]
+; AVX1-NEXT:    vpackusdw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm1
+; AVX1-NEXT:    vmovups %ymm1, (%rax)
+; AVX1-NEXT:    vmovups %ymm0, (%rax)
+; AVX1-NEXT:    vzeroupper
+; AVX1-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
@@ -1785,26 +2160,12 @@ define void @avg_v4i8_const(<4 x i8>* %a) {
 ; 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
+; AVX-LABEL: avg_v4i8_const:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX-NEXT:    vpavgb {{.*}}(%rip), %xmm0, %xmm0
+; AVX-NEXT:    vmovd %xmm0, (%rax)
+; AVX-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>
@@ -1822,26 +2183,12 @@ define void @avg_v8i8_const(<8 x i8>* %a) {
 ; 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
+; AVX-LABEL: avg_v8i8_const:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX-NEXT:    vpavgb {{.*}}(%rip), %xmm0, %xmm0
+; AVX-NEXT:    vmovq %xmm0, (%rax)
+; AVX-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>
@@ -1859,26 +2206,12 @@ define void @avg_v16i8_const(<16 x i8>* %a) {
 ; 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
+; AVX-LABEL: avg_v16i8_const:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX-NEXT:    vpavgb {{.*}}(%rip), %xmm0, %xmm0
+; AVX-NEXT:    vmovdqu %xmm0, (%rax)
+; AVX-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>
@@ -1949,6 +2282,54 @@ define void @avg_v32i8_const(<32 x i8>* %a) {
 ; SSE2-NEXT:    movdqu %xmm4, (%rax)
 ; SSE2-NEXT:    retq
 ;
+; AVX1-LABEL: avg_v32i8_const:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm8 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm6 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm0 = [5,6,7,8]
+; AVX1-NEXT:    vpaddd %xmm0, %xmm7, %xmm9
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm7 = [1,2,3,4]
+; AVX1-NEXT:    vpaddd %xmm7, %xmm6, %xmm6
+; AVX1-NEXT:    vpaddd %xmm0, %xmm5, %xmm5
+; AVX1-NEXT:    vpaddd %xmm7, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm0, %xmm3, %xmm3
+; AVX1-NEXT:    vpaddd %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddd %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpaddd %xmm7, %xmm8, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm1, %xmm8
+; AVX1-NEXT:    vpsrld $1, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm3
+; AVX1-NEXT:    vpsrld $1, %xmm4, %xmm4
+; AVX1-NEXT:    vpsrld $1, %xmm5, %xmm5
+; AVX1-NEXT:    vpsrld $1, %xmm6, %xmm6
+; AVX1-NEXT:    vpsrld $1, %xmm9, %xmm7
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
+; AVX1-NEXT:    vpand %xmm1, %xmm7, %xmm7
+; AVX1-NEXT:    vpand %xmm1, %xmm6, %xmm6
+; AVX1-NEXT:    vpackuswb %xmm7, %xmm6, %xmm6
+; AVX1-NEXT:    vpand %xmm1, %xmm5, %xmm5
+; AVX1-NEXT:    vpand %xmm1, %xmm4, %xmm4
+; AVX1-NEXT:    vpackuswb %xmm5, %xmm4, %xmm4
+; AVX1-NEXT:    vpackuswb %xmm6, %xmm4, %xmm4
+; AVX1-NEXT:    vpand %xmm1, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm1, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpand %xmm1, %xmm8, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm4, %ymm0, %ymm0
+; AVX1-NEXT:    vmovups %ymm0, (%rax)
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
 ; AVX2-LABEL: avg_v32i8_const:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vmovdqa (%rdi), %ymm0
@@ -1957,21 +2338,13 @@ define void @avg_v32i8_const(<32 x i8>* %a) {
 ; 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:    vzeroupper
-; 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:    vzeroupper
-; AVX512BW-NEXT:    retq
+; AVX512-LABEL: avg_v32i8_const:
+; AVX512:       # BB#0:
+; AVX512-NEXT:    vmovdqa (%rdi), %ymm0
+; AVX512-NEXT:    vpavgb {{.*}}(%rip), %ymm0, %ymm0
+; AVX512-NEXT:    vmovdqu %ymm0, (%rax)
+; AVX512-NEXT:    vzeroupper
+; AVX512-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>
@@ -2101,6 +2474,99 @@ define void @avg_v64i8_const(<64 x i8>* %a) {
 ; SSE2-NEXT:    movdqu %xmm1, (%rax)
 ; SSE2-NEXT:    retq
 ;
+; AVX1-LABEL: avg_v64i8_const:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vmovdqa %xmm0, -{{[0-9]+}}(%rsp) # 16-byte Spill
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm9 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm14 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm6 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm11 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm8 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm13 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm10 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm7 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm12 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm0 = [5,6,7,8]
+; AVX1-NEXT:    vpaddd %xmm0, %xmm5, %xmm5
+; AVX1-NEXT:    vpaddd %xmm0, %xmm7, %xmm15
+; AVX1-NEXT:    vpaddd %xmm0, %xmm13, %xmm13
+; AVX1-NEXT:    vpaddd %xmm0, %xmm1, %xmm7
+; AVX1-NEXT:    vpaddd %xmm0, %xmm11, %xmm11
+; AVX1-NEXT:    vpaddd %xmm0, %xmm3, %xmm1
+; AVX1-NEXT:    vmovdqa %xmm1, -{{[0-9]+}}(%rsp) # 16-byte Spill
+; AVX1-NEXT:    vpaddd %xmm0, %xmm9, %xmm9
+; AVX1-NEXT:    vpaddd %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vmovdqa %xmm0, -{{[0-9]+}}(%rsp) # 16-byte Spill
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [1,2,3,4]
+; AVX1-NEXT:    vpaddd %xmm2, %xmm12, %xmm0
+; AVX1-NEXT:    vpaddd %xmm2, %xmm10, %xmm10
+; AVX1-NEXT:    vpaddd %xmm2, %xmm8, %xmm8
+; AVX1-NEXT:    vpaddd %xmm2, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm2, %xmm6, %xmm1
+; AVX1-NEXT:    vpaddd %xmm2, %xmm14, %xmm6
+; AVX1-NEXT:    vpaddd -{{[0-9]+}}(%rsp), %xmm2, %xmm12 # 16-byte Folded Reload
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpaddd %xmm2, %xmm3, %xmm14
+; AVX1-NEXT:    vpsrld $1, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm5, %xmm3
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm5 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
+; AVX1-NEXT:    vpand %xmm5, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm5, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm10, %xmm3
+; AVX1-NEXT:    vpsrld $1, %xmm15, %xmm2
+; AVX1-NEXT:    vpand %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vpand %xmm5, %xmm3, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm8, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm13, %xmm3
+; AVX1-NEXT:    vpand %xmm5, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm4, %xmm3
+; AVX1-NEXT:    vpsrld $1, %xmm7, %xmm4
+; AVX1-NEXT:    vpand %xmm5, %xmm4, %xmm4
+; AVX1-NEXT:    vpand %xmm5, %xmm3, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm2, %ymm0
+; AVX1-NEXT:    vpsrld $1, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm11, %xmm2
+; AVX1-NEXT:    vpand %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vpand %xmm5, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm6, %xmm2
+; AVX1-NEXT:    vmovdqa -{{[0-9]+}}(%rsp), %xmm3 # 16-byte Reload
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm5, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm12, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm9, %xmm3
+; AVX1-NEXT:    vpand %xmm5, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm14, %xmm3
+; AVX1-NEXT:    vmovdqa -{{[0-9]+}}(%rsp), %xmm4 # 16-byte Reload
+; AVX1-NEXT:    vpsrld $1, %xmm4, %xmm4
+; AVX1-NEXT:    vpand %xmm5, %xmm4, %xmm4
+; AVX1-NEXT:    vpand %xmm5, %xmm3, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm1
+; AVX1-NEXT:    vmovups %ymm1, (%rax)
+; AVX1-NEXT:    vmovups %ymm0, (%rax)
+; AVX1-NEXT:    vzeroupper
+; AVX1-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
@@ -2215,26 +2681,12 @@ define void @avg_v4i16_const(<4 x i16>* %a) {
 ; 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
+; AVX-LABEL: avg_v4i16_const:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX-NEXT:    vpavgw {{.*}}(%rip), %xmm0, %xmm0
+; AVX-NEXT:    vmovq %xmm0, (%rax)
+; AVX-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>
@@ -2252,26 +2704,12 @@ define void @avg_v8i16_const(<8 x i16>* %a) {
 ; 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
+; AVX-LABEL: avg_v8i16_const:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX-NEXT:    vpavgw {{.*}}(%rip), %xmm0, %xmm0
+; AVX-NEXT:    vmovdqu %xmm0, (%rax)
+; AVX-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>
@@ -2317,6 +2755,34 @@ define void @avg_v16i16_const(<16 x i16>* %a) {
 ; SSE2-NEXT:    movdqu %xmm2, (%rax)
 ; SSE2-NEXT:    retq
 ;
+; AVX1-LABEL: avg_v16i16_const:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [5,6,7,8]
+; AVX1-NEXT:    vpaddd %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm5 = [1,2,3,4]
+; AVX1-NEXT:    vpaddd %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddd %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpaddd %xmm5, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm3 = xmm3[0],xmm4[1],xmm3[2],xmm4[3],xmm3[4],xmm4[5],xmm3[6],xmm4[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm2[0],xmm4[1],xmm2[2],xmm4[3],xmm2[4],xmm4[5],xmm2[6],xmm4[7]
+; AVX1-NEXT:    vpackusdw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0],xmm4[1],xmm1[2],xmm4[3],xmm1[4],xmm4[5],xmm1[6],xmm4[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],xmm4[1],xmm0[2],xmm4[3],xmm0[4],xmm4[5],xmm0[6],xmm4[7]
+; AVX1-NEXT:    vpackusdw %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vmovups %ymm0, (%rax)
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
 ; AVX2-LABEL: avg_v16i16_const:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vmovdqa (%rdi), %ymm0
@@ -2325,21 +2791,13 @@ define void @avg_v16i16_const(<16 x i16>* %a) {
 ; 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:    vzeroupper
-; 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:    vzeroupper
-; AVX512BW-NEXT:    retq
+; AVX512-LABEL: avg_v16i16_const:
+; AVX512:       # BB#0:
+; AVX512-NEXT:    vmovdqa (%rdi), %ymm0
+; AVX512-NEXT:    vpavgw {{.*}}(%rip), %ymm0, %ymm0
+; AVX512-NEXT:    vmovdqu %ymm0, (%rax)
+; AVX512-NEXT:    vzeroupper
+; AVX512-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>
@@ -2413,6 +2871,54 @@ define void @avg_v32i16_const(<32 x i16>* %a) {
 ; SSE2-NEXT:    movdqu %xmm5, (%rax)
 ; SSE2-NEXT:    retq
 ;
+; AVX1-LABEL: avg_v32i16_const:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm8 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm4 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm5 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm6 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm7 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm0 = [5,6,7,8]
+; AVX1-NEXT:    vpaddd %xmm0, %xmm7, %xmm9
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm7 = [1,2,3,4]
+; AVX1-NEXT:    vpaddd %xmm7, %xmm6, %xmm6
+; AVX1-NEXT:    vpaddd %xmm0, %xmm5, %xmm5
+; AVX1-NEXT:    vpaddd %xmm7, %xmm4, %xmm4
+; AVX1-NEXT:    vpaddd %xmm0, %xmm3, %xmm3
+; AVX1-NEXT:    vpaddd %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddd %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpaddd %xmm7, %xmm8, %xmm1
+; AVX1-NEXT:    vpsrld $1, %xmm1, %xmm8
+; AVX1-NEXT:    vpsrld $1, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrld $1, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm3
+; AVX1-NEXT:    vpsrld $1, %xmm4, %xmm4
+; AVX1-NEXT:    vpsrld $1, %xmm5, %xmm5
+; AVX1-NEXT:    vpsrld $1, %xmm6, %xmm6
+; AVX1-NEXT:    vpsrld $1, %xmm9, %xmm7
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm7 = xmm7[0],xmm1[1],xmm7[2],xmm1[3],xmm7[4],xmm1[5],xmm7[6],xmm1[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm6 = xmm6[0],xmm1[1],xmm6[2],xmm1[3],xmm6[4],xmm1[5],xmm6[6],xmm1[7]
+; AVX1-NEXT:    vpackusdw %xmm7, %xmm6, %xmm6
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm5 = xmm5[0],xmm1[1],xmm5[2],xmm1[3],xmm5[4],xmm1[5],xmm5[6],xmm1[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm4 = xmm4[0],xmm1[1],xmm4[2],xmm1[3],xmm4[4],xmm1[5],xmm4[6],xmm1[7]
+; AVX1-NEXT:    vpackusdw %xmm5, %xmm4, %xmm4
+; AVX1-NEXT:    vinsertf128 $1, %xmm6, %ymm4, %ymm4
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm3 = xmm3[0],xmm1[1],xmm3[2],xmm1[3],xmm3[4],xmm1[5],xmm3[6],xmm1[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm2[0],xmm1[1],xmm2[2],xmm1[3],xmm2[4],xmm1[5],xmm2[6],xmm1[7]
+; AVX1-NEXT:    vpackusdw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3],xmm0[4],xmm1[5],xmm0[6],xmm1[7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm8[0],xmm1[1],xmm8[2],xmm1[3],xmm8[4],xmm1[5],xmm8[6],xmm1[7]
+; AVX1-NEXT:    vpackusdw %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vmovups %ymm0, (%rax)
+; AVX1-NEXT:    vmovups %ymm4, (%rax)
+; AVX1-NEXT:    vzeroupper
+; AVX1-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