summaryrefslogtreecommitdiff
path: root/test/CodeGen/X86/vector-tzcnt-256.ll
diff options
context:
space:
mode:
Diffstat (limited to 'test/CodeGen/X86/vector-tzcnt-256.ll')
-rw-r--r--test/CodeGen/X86/vector-tzcnt-256.ll379
1 files changed, 243 insertions, 136 deletions
diff --git a/test/CodeGen/X86/vector-tzcnt-256.ll b/test/CodeGen/X86/vector-tzcnt-256.ll
index a0b277ddd7327..101ae95550e72 100644
--- a/test/CodeGen/X86/vector-tzcnt-256.ll
+++ b/test/CodeGen/X86/vector-tzcnt-256.ll
@@ -3,6 +3,7 @@
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512cd,+avx512vl | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX512CDVL
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512cd,-avx512vl | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX512CD
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vpopcntdq | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX512VPOPCNTDQ
;
; Just one 32-bit run to make sure we do reasonable things for i64 tzcnt.
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=X32-AVX --check-prefix=X32-AVX2
@@ -12,11 +13,8 @@ define <4 x i64> @testv4i64(<4 x i64> %in) nounwind {
; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT: vpsubq %xmm1, %xmm2, %xmm1
-; AVX1-NEXT: vpsubq %xmm0, %xmm2, %xmm3
-; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm3, %ymm1
-; AVX1-NEXT: vandps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT: vpsubq %xmm1, %xmm2, %xmm3
+; AVX1-NEXT: vpand %xmm3, %xmm1, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [1,1]
; AVX1-NEXT: vpsubq %xmm3, %xmm1, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
@@ -28,6 +26,8 @@ define <4 x i64> @testv4i64(<4 x i64> %in) nounwind {
; AVX1-NEXT: vpshufb %xmm1, %xmm6, %xmm1
; AVX1-NEXT: vpaddb %xmm5, %xmm1, %xmm1
; AVX1-NEXT: vpsadbw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT: vpsubq %xmm0, %xmm2, %xmm5
+; AVX1-NEXT: vpand %xmm5, %xmm0, %xmm0
; AVX1-NEXT: vpsubq %xmm3, %xmm0, %xmm0
; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm3
; AVX1-NEXT: vpshufb %xmm3, %xmm6, %xmm3
@@ -92,6 +92,17 @@ define <4 x i64> @testv4i64(<4 x i64> %in) nounwind {
; AVX512CD-NEXT: vpsadbw %ymm1, %ymm0, %ymm0
; AVX512CD-NEXT: retq
;
+; AVX512VPOPCNTDQ-LABEL: testv4i64:
+; AVX512VPOPCNTDQ: # BB#0:
+; AVX512VPOPCNTDQ-NEXT: vpxor %ymm1, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpsubq %ymm0, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpbroadcastq {{.*}}(%rip), %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpsubq %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpopcntq %zmm0, %zmm0
+; AVX512VPOPCNTDQ-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<kill>
+; AVX512VPOPCNTDQ-NEXT: retq
+;
; X32-AVX-LABEL: testv4i64:
; X32-AVX: # BB#0:
; X32-AVX-NEXT: vpxor %ymm1, %ymm1, %ymm1
@@ -117,11 +128,8 @@ define <4 x i64> @testv4i64u(<4 x i64> %in) nounwind {
; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT: vpsubq %xmm1, %xmm2, %xmm1
-; AVX1-NEXT: vpsubq %xmm0, %xmm2, %xmm3
-; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm3, %ymm1
-; AVX1-NEXT: vandps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT: vpsubq %xmm1, %xmm2, %xmm3
+; AVX1-NEXT: vpand %xmm3, %xmm1, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [1,1]
; AVX1-NEXT: vpsubq %xmm3, %xmm1, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
@@ -133,6 +141,8 @@ define <4 x i64> @testv4i64u(<4 x i64> %in) nounwind {
; AVX1-NEXT: vpshufb %xmm1, %xmm6, %xmm1
; AVX1-NEXT: vpaddb %xmm5, %xmm1, %xmm1
; AVX1-NEXT: vpsadbw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT: vpsubq %xmm0, %xmm2, %xmm5
+; AVX1-NEXT: vpand %xmm5, %xmm0, %xmm0
; AVX1-NEXT: vpsubq %xmm3, %xmm0, %xmm0
; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm3
; AVX1-NEXT: vpshufb %xmm3, %xmm6, %xmm3
@@ -182,6 +192,17 @@ define <4 x i64> @testv4i64u(<4 x i64> %in) nounwind {
; AVX512CD-NEXT: vpsubq %ymm0, %ymm1, %ymm0
; AVX512CD-NEXT: retq
;
+; AVX512VPOPCNTDQ-LABEL: testv4i64u:
+; AVX512VPOPCNTDQ: # BB#0:
+; AVX512VPOPCNTDQ-NEXT: vpxor %ymm1, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpsubq %ymm0, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpbroadcastq {{.*}}(%rip), %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpsubq %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpopcntq %zmm0, %zmm0
+; AVX512VPOPCNTDQ-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<kill>
+; AVX512VPOPCNTDQ-NEXT: retq
+;
; X32-AVX-LABEL: testv4i64u:
; X32-AVX: # BB#0:
; X32-AVX-NEXT: vpxor %ymm1, %ymm1, %ymm1
@@ -205,28 +226,27 @@ define <4 x i64> @testv4i64u(<4 x i64> %in) nounwind {
define <8 x i32> @testv8i32(<8 x i32> %in) nounwind {
; AVX1-LABEL: testv8i32:
; AVX1: # BB#0:
-; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
-; AVX1-NEXT: vpxor %xmm1, %xmm1, %xmm1
-; AVX1-NEXT: vpsubd %xmm2, %xmm1, %xmm2
-; AVX1-NEXT: vpsubd %xmm0, %xmm1, %xmm3
-; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm3, %ymm2
-; AVX1-NEXT: vandps %ymm2, %ymm0, %ymm0
-; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT: vpsubd %xmm1, %xmm2, %xmm3
+; AVX1-NEXT: vpand %xmm3, %xmm1, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [1,1,1,1]
-; AVX1-NEXT: vpsubd %xmm3, %xmm2, %xmm2
+; AVX1-NEXT: vpsubd %xmm3, %xmm1, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; AVX1-NEXT: vpand %xmm4, %xmm2, %xmm5
+; AVX1-NEXT: vpand %xmm4, %xmm1, %xmm5
; AVX1-NEXT: vmovdqa {{.*#+}} xmm6 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
; AVX1-NEXT: vpshufb %xmm5, %xmm6, %xmm5
-; AVX1-NEXT: vpsrlw $4, %xmm2, %xmm2
-; AVX1-NEXT: vpand %xmm4, %xmm2, %xmm2
-; AVX1-NEXT: vpshufb %xmm2, %xmm6, %xmm2
-; AVX1-NEXT: vpaddb %xmm5, %xmm2, %xmm2
-; AVX1-NEXT: vpunpckhdq {{.*#+}} xmm5 = xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; AVX1-NEXT: vpsadbw %xmm1, %xmm5, %xmm5
-; AVX1-NEXT: vpmovzxdq {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero
-; AVX1-NEXT: vpsadbw %xmm1, %xmm2, %xmm2
-; AVX1-NEXT: vpackuswb %xmm5, %xmm2, %xmm2
+; AVX1-NEXT: vpsrlw $4, %xmm1, %xmm1
+; AVX1-NEXT: vpand %xmm4, %xmm1, %xmm1
+; AVX1-NEXT: vpshufb %xmm1, %xmm6, %xmm1
+; AVX1-NEXT: vpaddb %xmm5, %xmm1, %xmm1
+; AVX1-NEXT: vpunpckhdq {{.*#+}} xmm5 = xmm1[2],xmm2[2],xmm1[3],xmm2[3]
+; AVX1-NEXT: vpsadbw %xmm2, %xmm5, %xmm5
+; AVX1-NEXT: vpmovzxdq {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero
+; AVX1-NEXT: vpsadbw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT: vpackuswb %xmm5, %xmm1, %xmm1
+; AVX1-NEXT: vpsubd %xmm0, %xmm2, %xmm5
+; AVX1-NEXT: vpand %xmm5, %xmm0, %xmm0
; AVX1-NEXT: vpsubd %xmm3, %xmm0, %xmm0
; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm3
; AVX1-NEXT: vpshufb %xmm3, %xmm6, %xmm3
@@ -234,12 +254,12 @@ define <8 x i32> @testv8i32(<8 x i32> %in) nounwind {
; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm0
; AVX1-NEXT: vpshufb %xmm0, %xmm6, %xmm0
; AVX1-NEXT: vpaddb %xmm3, %xmm0, %xmm0
-; AVX1-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT: vpsadbw %xmm1, %xmm3, %xmm3
+; AVX1-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
+; AVX1-NEXT: vpsadbw %xmm2, %xmm3, %xmm3
; AVX1-NEXT: vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
-; AVX1-NEXT: vpsadbw %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpsadbw %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vpackuswb %xmm3, %xmm0, %xmm0
-; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: testv8i32:
@@ -307,6 +327,17 @@ define <8 x i32> @testv8i32(<8 x i32> %in) nounwind {
; AVX512CD-NEXT: vpackuswb %ymm2, %ymm0, %ymm0
; AVX512CD-NEXT: retq
;
+; AVX512VPOPCNTDQ-LABEL: testv8i32:
+; AVX512VPOPCNTDQ: # BB#0:
+; AVX512VPOPCNTDQ-NEXT: vpxor %ymm1, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpsubd %ymm0, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpbroadcastd {{.*}}(%rip), %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpsubd %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpopcntd %zmm0, %zmm0
+; AVX512VPOPCNTDQ-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<kill>
+; AVX512VPOPCNTDQ-NEXT: retq
+;
; X32-AVX-LABEL: testv8i32:
; X32-AVX: # BB#0:
; X32-AVX-NEXT: vpxor %ymm1, %ymm1, %ymm1
@@ -335,28 +366,27 @@ define <8 x i32> @testv8i32(<8 x i32> %in) nounwind {
define <8 x i32> @testv8i32u(<8 x i32> %in) nounwind {
; AVX1-LABEL: testv8i32u:
; AVX1: # BB#0:
-; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
-; AVX1-NEXT: vpxor %xmm1, %xmm1, %xmm1
-; AVX1-NEXT: vpsubd %xmm2, %xmm1, %xmm2
-; AVX1-NEXT: vpsubd %xmm0, %xmm1, %xmm3
-; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm3, %ymm2
-; AVX1-NEXT: vandps %ymm2, %ymm0, %ymm0
-; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT: vpsubd %xmm1, %xmm2, %xmm3
+; AVX1-NEXT: vpand %xmm3, %xmm1, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [1,1,1,1]
-; AVX1-NEXT: vpsubd %xmm3, %xmm2, %xmm2
+; AVX1-NEXT: vpsubd %xmm3, %xmm1, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; AVX1-NEXT: vpand %xmm4, %xmm2, %xmm5
+; AVX1-NEXT: vpand %xmm4, %xmm1, %xmm5
; AVX1-NEXT: vmovdqa {{.*#+}} xmm6 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
; AVX1-NEXT: vpshufb %xmm5, %xmm6, %xmm5
-; AVX1-NEXT: vpsrlw $4, %xmm2, %xmm2
-; AVX1-NEXT: vpand %xmm4, %xmm2, %xmm2
-; AVX1-NEXT: vpshufb %xmm2, %xmm6, %xmm2
-; AVX1-NEXT: vpaddb %xmm5, %xmm2, %xmm2
-; AVX1-NEXT: vpunpckhdq {{.*#+}} xmm5 = xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; AVX1-NEXT: vpsadbw %xmm1, %xmm5, %xmm5
-; AVX1-NEXT: vpmovzxdq {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero
-; AVX1-NEXT: vpsadbw %xmm1, %xmm2, %xmm2
-; AVX1-NEXT: vpackuswb %xmm5, %xmm2, %xmm2
+; AVX1-NEXT: vpsrlw $4, %xmm1, %xmm1
+; AVX1-NEXT: vpand %xmm4, %xmm1, %xmm1
+; AVX1-NEXT: vpshufb %xmm1, %xmm6, %xmm1
+; AVX1-NEXT: vpaddb %xmm5, %xmm1, %xmm1
+; AVX1-NEXT: vpunpckhdq {{.*#+}} xmm5 = xmm1[2],xmm2[2],xmm1[3],xmm2[3]
+; AVX1-NEXT: vpsadbw %xmm2, %xmm5, %xmm5
+; AVX1-NEXT: vpmovzxdq {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero
+; AVX1-NEXT: vpsadbw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT: vpackuswb %xmm5, %xmm1, %xmm1
+; AVX1-NEXT: vpsubd %xmm0, %xmm2, %xmm5
+; AVX1-NEXT: vpand %xmm5, %xmm0, %xmm0
; AVX1-NEXT: vpsubd %xmm3, %xmm0, %xmm0
; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm3
; AVX1-NEXT: vpshufb %xmm3, %xmm6, %xmm3
@@ -364,12 +394,12 @@ define <8 x i32> @testv8i32u(<8 x i32> %in) nounwind {
; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm0
; AVX1-NEXT: vpshufb %xmm0, %xmm6, %xmm0
; AVX1-NEXT: vpaddb %xmm3, %xmm0, %xmm0
-; AVX1-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT: vpsadbw %xmm1, %xmm3, %xmm3
+; AVX1-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm0[2],xmm2[2],xmm0[3],xmm2[3]
+; AVX1-NEXT: vpsadbw %xmm2, %xmm3, %xmm3
; AVX1-NEXT: vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
-; AVX1-NEXT: vpsadbw %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpsadbw %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vpackuswb %xmm3, %xmm0, %xmm0
-; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: testv8i32u:
@@ -414,6 +444,17 @@ define <8 x i32> @testv8i32u(<8 x i32> %in) nounwind {
; AVX512CD-NEXT: vpsubd %ymm0, %ymm1, %ymm0
; AVX512CD-NEXT: retq
;
+; AVX512VPOPCNTDQ-LABEL: testv8i32u:
+; AVX512VPOPCNTDQ: # BB#0:
+; AVX512VPOPCNTDQ-NEXT: vpxor %ymm1, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpsubd %ymm0, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpbroadcastd {{.*}}(%rip), %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpsubd %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpopcntd %zmm0, %zmm0
+; AVX512VPOPCNTDQ-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<kill>
+; AVX512VPOPCNTDQ-NEXT: retq
+;
; X32-AVX-LABEL: testv8i32u:
; X32-AVX: # BB#0:
; X32-AVX-NEXT: vpxor %ymm1, %ymm1, %ymm1
@@ -442,32 +483,31 @@ define <8 x i32> @testv8i32u(<8 x i32> %in) nounwind {
define <16 x i16> @testv16i16(<16 x i16> %in) nounwind {
; AVX1-LABEL: testv16i16:
; AVX1: # BB#0:
-; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT: vpsubw %xmm1, %xmm2, %xmm1
-; AVX1-NEXT: vpsubw %xmm0, %xmm2, %xmm2
-; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm2, %ymm1
-; AVX1-NEXT: vandps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm1 = [1,1,1,1,1,1,1,1]
-; AVX1-NEXT: vpsubw %xmm1, %xmm0, %xmm2
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; AVX1-NEXT: vpand %xmm3, %xmm2, %xmm4
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm5 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
-; AVX1-NEXT: vpshufb %xmm4, %xmm5, %xmm4
+; AVX1-NEXT: vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT: vpsubw %xmm0, %xmm1, %xmm2
+; AVX1-NEXT: vpand %xmm2, %xmm0, %xmm2
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [1,1,1,1,1,1,1,1]
+; AVX1-NEXT: vpsubw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
+; AVX1-NEXT: vpand %xmm4, %xmm2, %xmm5
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm6 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
+; AVX1-NEXT: vpshufb %xmm5, %xmm6, %xmm5
; AVX1-NEXT: vpsrlw $4, %xmm2, %xmm2
-; AVX1-NEXT: vpand %xmm3, %xmm2, %xmm2
-; AVX1-NEXT: vpshufb %xmm2, %xmm5, %xmm2
-; AVX1-NEXT: vpaddb %xmm4, %xmm2, %xmm2
-; AVX1-NEXT: vpsllw $8, %xmm2, %xmm4
-; AVX1-NEXT: vpaddb %xmm2, %xmm4, %xmm2
+; AVX1-NEXT: vpand %xmm4, %xmm2, %xmm2
+; AVX1-NEXT: vpshufb %xmm2, %xmm6, %xmm2
+; AVX1-NEXT: vpaddb %xmm5, %xmm2, %xmm2
+; AVX1-NEXT: vpsllw $8, %xmm2, %xmm5
+; AVX1-NEXT: vpaddb %xmm2, %xmm5, %xmm2
; AVX1-NEXT: vpsrlw $8, %xmm2, %xmm2
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
-; AVX1-NEXT: vpsubw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT: vpand %xmm3, %xmm0, %xmm1
-; AVX1-NEXT: vpshufb %xmm1, %xmm5, %xmm1
+; AVX1-NEXT: vpsubw %xmm0, %xmm1, %xmm1
+; AVX1-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpsubw %xmm3, %xmm0, %xmm0
+; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm1
+; AVX1-NEXT: vpshufb %xmm1, %xmm6, %xmm1
; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm0
-; AVX1-NEXT: vpand %xmm3, %xmm0, %xmm0
-; AVX1-NEXT: vpshufb %xmm0, %xmm5, %xmm0
+; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm0
+; AVX1-NEXT: vpshufb %xmm0, %xmm6, %xmm0
; AVX1-NEXT: vpaddb %xmm1, %xmm0, %xmm0
; AVX1-NEXT: vpsllw $8, %xmm0, %xmm1
; AVX1-NEXT: vpaddb %xmm0, %xmm1, %xmm0
@@ -532,6 +572,25 @@ define <16 x i16> @testv16i16(<16 x i16> %in) nounwind {
; AVX512CD-NEXT: vpsrlw $8, %ymm0, %ymm0
; AVX512CD-NEXT: retq
;
+; AVX512VPOPCNTDQ-LABEL: testv16i16:
+; AVX512VPOPCNTDQ: # BB#0:
+; AVX512VPOPCNTDQ-NEXT: vpxor %ymm1, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpsubw %ymm0, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpsubw {{.*}}(%rip), %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vmovdqa {{.*#+}} ymm1 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm2
+; AVX512VPOPCNTDQ-NEXT: vmovdqa {{.*#+}} ymm3 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4,0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
+; AVX512VPOPCNTDQ-NEXT: vpshufb %ymm2, %ymm3, %ymm2
+; AVX512VPOPCNTDQ-NEXT: vpsrlw $4, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpshufb %ymm0, %ymm3, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpaddb %ymm2, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpsllw $8, %ymm0, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpaddb %ymm0, %ymm1, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpsrlw $8, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: retq
+;
; X32-AVX-LABEL: testv16i16:
; X32-AVX: # BB#0:
; X32-AVX-NEXT: vpxor %ymm1, %ymm1, %ymm1
@@ -557,32 +616,31 @@ define <16 x i16> @testv16i16(<16 x i16> %in) nounwind {
define <16 x i16> @testv16i16u(<16 x i16> %in) nounwind {
; AVX1-LABEL: testv16i16u:
; AVX1: # BB#0:
-; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT: vpsubw %xmm1, %xmm2, %xmm1
-; AVX1-NEXT: vpsubw %xmm0, %xmm2, %xmm2
-; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm2, %ymm1
-; AVX1-NEXT: vandps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm1 = [1,1,1,1,1,1,1,1]
-; AVX1-NEXT: vpsubw %xmm1, %xmm0, %xmm2
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; AVX1-NEXT: vpand %xmm3, %xmm2, %xmm4
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm5 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
-; AVX1-NEXT: vpshufb %xmm4, %xmm5, %xmm4
+; AVX1-NEXT: vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT: vpsubw %xmm0, %xmm1, %xmm2
+; AVX1-NEXT: vpand %xmm2, %xmm0, %xmm2
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [1,1,1,1,1,1,1,1]
+; AVX1-NEXT: vpsubw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
+; AVX1-NEXT: vpand %xmm4, %xmm2, %xmm5
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm6 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
+; AVX1-NEXT: vpshufb %xmm5, %xmm6, %xmm5
; AVX1-NEXT: vpsrlw $4, %xmm2, %xmm2
-; AVX1-NEXT: vpand %xmm3, %xmm2, %xmm2
-; AVX1-NEXT: vpshufb %xmm2, %xmm5, %xmm2
-; AVX1-NEXT: vpaddb %xmm4, %xmm2, %xmm2
-; AVX1-NEXT: vpsllw $8, %xmm2, %xmm4
-; AVX1-NEXT: vpaddb %xmm2, %xmm4, %xmm2
+; AVX1-NEXT: vpand %xmm4, %xmm2, %xmm2
+; AVX1-NEXT: vpshufb %xmm2, %xmm6, %xmm2
+; AVX1-NEXT: vpaddb %xmm5, %xmm2, %xmm2
+; AVX1-NEXT: vpsllw $8, %xmm2, %xmm5
+; AVX1-NEXT: vpaddb %xmm2, %xmm5, %xmm2
; AVX1-NEXT: vpsrlw $8, %xmm2, %xmm2
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm0
-; AVX1-NEXT: vpsubw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT: vpand %xmm3, %xmm0, %xmm1
-; AVX1-NEXT: vpshufb %xmm1, %xmm5, %xmm1
+; AVX1-NEXT: vpsubw %xmm0, %xmm1, %xmm1
+; AVX1-NEXT: vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT: vpsubw %xmm3, %xmm0, %xmm0
+; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm1
+; AVX1-NEXT: vpshufb %xmm1, %xmm6, %xmm1
; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm0
-; AVX1-NEXT: vpand %xmm3, %xmm0, %xmm0
-; AVX1-NEXT: vpshufb %xmm0, %xmm5, %xmm0
+; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm0
+; AVX1-NEXT: vpshufb %xmm0, %xmm6, %xmm0
; AVX1-NEXT: vpaddb %xmm1, %xmm0, %xmm0
; AVX1-NEXT: vpsllw $8, %xmm0, %xmm1
; AVX1-NEXT: vpaddb %xmm0, %xmm1, %xmm0
@@ -647,6 +705,25 @@ define <16 x i16> @testv16i16u(<16 x i16> %in) nounwind {
; AVX512CD-NEXT: vpsrlw $8, %ymm0, %ymm0
; AVX512CD-NEXT: retq
;
+; AVX512VPOPCNTDQ-LABEL: testv16i16u:
+; AVX512VPOPCNTDQ: # BB#0:
+; AVX512VPOPCNTDQ-NEXT: vpxor %ymm1, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpsubw %ymm0, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpsubw {{.*}}(%rip), %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vmovdqa {{.*#+}} ymm1 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm2
+; AVX512VPOPCNTDQ-NEXT: vmovdqa {{.*#+}} ymm3 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4,0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
+; AVX512VPOPCNTDQ-NEXT: vpshufb %ymm2, %ymm3, %ymm2
+; AVX512VPOPCNTDQ-NEXT: vpsrlw $4, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpshufb %ymm0, %ymm3, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpaddb %ymm2, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpsllw $8, %ymm0, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpaddb %ymm0, %ymm1, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpsrlw $8, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: retq
+;
; X32-AVX-LABEL: testv16i16u:
; X32-AVX: # BB#0:
; X32-AVX-NEXT: vpxor %ymm1, %ymm1, %ymm1
@@ -674,27 +751,26 @@ define <32 x i8> @testv32i8(<32 x i8> %in) nounwind {
; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT: vpsubb %xmm1, %xmm2, %xmm1
-; AVX1-NEXT: vpsubb %xmm0, %xmm2, %xmm2
-; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm2, %ymm1
-; AVX1-NEXT: vandps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
-; AVX1-NEXT: vpsubb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; AVX1-NEXT: vpand %xmm3, %xmm1, %xmm4
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm5 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
-; AVX1-NEXT: vpshufb %xmm4, %xmm5, %xmm4
-; AVX1-NEXT: vpsrlw $4, %xmm1, %xmm1
+; AVX1-NEXT: vpsubb %xmm1, %xmm2, %xmm3
; AVX1-NEXT: vpand %xmm3, %xmm1, %xmm1
-; AVX1-NEXT: vpshufb %xmm1, %xmm5, %xmm1
-; AVX1-NEXT: vpaddb %xmm4, %xmm1, %xmm1
-; AVX1-NEXT: vpsubb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT: vpand %xmm3, %xmm0, %xmm2
-; AVX1-NEXT: vpshufb %xmm2, %xmm5, %xmm2
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX1-NEXT: vpsubb %xmm3, %xmm1, %xmm1
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
+; AVX1-NEXT: vpand %xmm4, %xmm1, %xmm5
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm6 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
+; AVX1-NEXT: vpshufb %xmm5, %xmm6, %xmm5
+; AVX1-NEXT: vpsrlw $4, %xmm1, %xmm1
+; AVX1-NEXT: vpand %xmm4, %xmm1, %xmm1
+; AVX1-NEXT: vpshufb %xmm1, %xmm6, %xmm1
+; AVX1-NEXT: vpaddb %xmm5, %xmm1, %xmm1
+; AVX1-NEXT: vpsubb %xmm0, %xmm2, %xmm2
+; AVX1-NEXT: vpand %xmm2, %xmm0, %xmm0
+; AVX1-NEXT: vpsubb %xmm3, %xmm0, %xmm0
+; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm2
+; AVX1-NEXT: vpshufb %xmm2, %xmm6, %xmm2
; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm0
-; AVX1-NEXT: vpand %xmm3, %xmm0, %xmm0
-; AVX1-NEXT: vpshufb %xmm0, %xmm5, %xmm0
+; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm0
+; AVX1-NEXT: vpshufb %xmm0, %xmm6, %xmm0
; AVX1-NEXT: vpaddb %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq
@@ -747,6 +823,22 @@ define <32 x i8> @testv32i8(<32 x i8> %in) nounwind {
; AVX512CD-NEXT: vpaddb %ymm2, %ymm0, %ymm0
; AVX512CD-NEXT: retq
;
+; AVX512VPOPCNTDQ-LABEL: testv32i8:
+; AVX512VPOPCNTDQ: # BB#0:
+; AVX512VPOPCNTDQ-NEXT: vpxor %ymm1, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpsubb %ymm0, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpsubb {{.*}}(%rip), %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vmovdqa {{.*#+}} ymm1 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm2
+; AVX512VPOPCNTDQ-NEXT: vmovdqa {{.*#+}} ymm3 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4,0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
+; AVX512VPOPCNTDQ-NEXT: vpshufb %ymm2, %ymm3, %ymm2
+; AVX512VPOPCNTDQ-NEXT: vpsrlw $4, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpshufb %ymm0, %ymm3, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpaddb %ymm2, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: retq
+;
; X32-AVX-LABEL: testv32i8:
; X32-AVX: # BB#0:
; X32-AVX-NEXT: vpxor %ymm1, %ymm1, %ymm1
@@ -771,27 +863,26 @@ define <32 x i8> @testv32i8u(<32 x i8> %in) nounwind {
; AVX1: # BB#0:
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT: vpsubb %xmm1, %xmm2, %xmm1
-; AVX1-NEXT: vpsubb %xmm0, %xmm2, %xmm2
-; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm2, %ymm1
-; AVX1-NEXT: vandps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
-; AVX1-NEXT: vpsubb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
-; AVX1-NEXT: vpand %xmm3, %xmm1, %xmm4
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm5 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
-; AVX1-NEXT: vpshufb %xmm4, %xmm5, %xmm4
-; AVX1-NEXT: vpsrlw $4, %xmm1, %xmm1
+; AVX1-NEXT: vpsubb %xmm1, %xmm2, %xmm3
; AVX1-NEXT: vpand %xmm3, %xmm1, %xmm1
-; AVX1-NEXT: vpshufb %xmm1, %xmm5, %xmm1
-; AVX1-NEXT: vpaddb %xmm4, %xmm1, %xmm1
-; AVX1-NEXT: vpsubb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT: vpand %xmm3, %xmm0, %xmm2
-; AVX1-NEXT: vpshufb %xmm2, %xmm5, %xmm2
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX1-NEXT: vpsubb %xmm3, %xmm1, %xmm1
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
+; AVX1-NEXT: vpand %xmm4, %xmm1, %xmm5
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm6 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
+; AVX1-NEXT: vpshufb %xmm5, %xmm6, %xmm5
+; AVX1-NEXT: vpsrlw $4, %xmm1, %xmm1
+; AVX1-NEXT: vpand %xmm4, %xmm1, %xmm1
+; AVX1-NEXT: vpshufb %xmm1, %xmm6, %xmm1
+; AVX1-NEXT: vpaddb %xmm5, %xmm1, %xmm1
+; AVX1-NEXT: vpsubb %xmm0, %xmm2, %xmm2
+; AVX1-NEXT: vpand %xmm2, %xmm0, %xmm0
+; AVX1-NEXT: vpsubb %xmm3, %xmm0, %xmm0
+; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm2
+; AVX1-NEXT: vpshufb %xmm2, %xmm6, %xmm2
; AVX1-NEXT: vpsrlw $4, %xmm0, %xmm0
-; AVX1-NEXT: vpand %xmm3, %xmm0, %xmm0
-; AVX1-NEXT: vpshufb %xmm0, %xmm5, %xmm0
+; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm0
+; AVX1-NEXT: vpshufb %xmm0, %xmm6, %xmm0
; AVX1-NEXT: vpaddb %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT: retq
@@ -844,6 +935,22 @@ define <32 x i8> @testv32i8u(<32 x i8> %in) nounwind {
; AVX512CD-NEXT: vpaddb %ymm2, %ymm0, %ymm0
; AVX512CD-NEXT: retq
;
+; AVX512VPOPCNTDQ-LABEL: testv32i8u:
+; AVX512VPOPCNTDQ: # BB#0:
+; AVX512VPOPCNTDQ-NEXT: vpxor %ymm1, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpsubb %ymm0, %ymm1, %ymm1
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpsubb {{.*}}(%rip), %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vmovdqa {{.*#+}} ymm1 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm2
+; AVX512VPOPCNTDQ-NEXT: vmovdqa {{.*#+}} ymm3 = [0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4,0,1,1,2,1,2,2,3,1,2,2,3,2,3,3,4]
+; AVX512VPOPCNTDQ-NEXT: vpshufb %ymm2, %ymm3, %ymm2
+; AVX512VPOPCNTDQ-NEXT: vpsrlw $4, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpand %ymm1, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpshufb %ymm0, %ymm3, %ymm0
+; AVX512VPOPCNTDQ-NEXT: vpaddb %ymm2, %ymm0, %ymm0
+; AVX512VPOPCNTDQ-NEXT: retq
+;
; X32-AVX-LABEL: testv32i8u:
; X32-AVX: # BB#0:
; X32-AVX-NEXT: vpxor %ymm1, %ymm1, %ymm1
generated by cgit v1.2.3 (git 2.25.1) at 2025-08-02 03:44:36 +0000