[X86] Add test files for upgrade of vbmi2 expand load and compress store intrinsics that was done in r334381.

llvm-svn: 334386

2 files changed, 651 insertions, 0 deletions

diff --git a/llvm/test/CodeGen/X86/avx512vbmi2-intrinsics-upgrade.ll b/llvm/test/CodeGen/X86/avx512vbmi2-intrinsics-upgrade.ll
new file mode 100644
index 00000000000..e1b11292418
--- /dev/null
+++ b/llvm/test/CodeGen/X86/avx512vbmi2-intrinsics-upgrade.ll
@@ -0,0 +1,281 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512vbmi2 --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X86
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vbmi2 --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X64
+
+define <32 x i16> @test_mask_expand_load_w_512(i8* %addr, <32 x i16> %data, i32 %mask) {
+; X86-LABEL: test_mask_expand_load_w_512:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpexpandw (%eax), %zmm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_expand_load_w_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpexpandw (%rdi), %zmm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <32 x i16> @llvm.x86.avx512.mask.expand.load.w.512(i8* %addr, <32 x i16> %data, i32 %mask)
+  ret <32 x i16> %res
+}
+
+define <32 x i16> @test_maskz_expand_load_w_512(i8* %addr, i32 %mask) {
+; X86-LABEL: test_maskz_expand_load_w_512:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpexpandw (%eax), %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xc9,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_maskz_expand_load_w_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpexpandw (%rdi), %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xc9,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <32 x i16> @llvm.x86.avx512.mask.expand.load.w.512(i8* %addr, <32 x i16> zeroinitializer, i32 %mask)
+  ret <32 x i16> %res
+}
+
+declare <32 x i16> @llvm.x86.avx512.mask.expand.load.w.512(i8* %addr, <32 x i16> %data, i32 %mask)
+
+define <32 x i16> @test_expand_load_w_512(i8* %addr, <32 x i16> %data) {
+; X86-LABEL: test_expand_load_w_512:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kxnord %k0, %k0, %k1 # encoding: [0xc4,0xe1,0xfd,0x46,0xc8]
+; X86-NEXT:    vpexpandw (%eax), %zmm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_expand_load_w_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kxnord %k0, %k0, %k1 # encoding: [0xc4,0xe1,0xfd,0x46,0xc8]
+; X64-NEXT:    vpexpandw (%rdi), %zmm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <32 x i16> @llvm.x86.avx512.mask.expand.load.w.512(i8* %addr, <32 x i16> %data, i32 -1)
+  ret <32 x i16> %res
+}
+
+define <64 x i8> @test_mask_expand_load_b_512(i8* %addr, <64 x i8> %data, i64 %mask) {
+; X86-LABEL: test_mask_expand_load_b_512:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovq {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf8,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpexpandb (%eax), %zmm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_expand_load_b_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovq %rsi, %k1 # encoding: [0xc4,0xe1,0xfb,0x92,0xce]
+; X64-NEXT:    vpexpandb (%rdi), %zmm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <64 x i8> @llvm.x86.avx512.mask.expand.load.b.512(i8* %addr, <64 x i8> %data, i64 %mask)
+  ret <64 x i8> %res
+}
+
+define <64 x i8> @test_maskz_expand_load_b_512(i8* %addr, i64 %mask) {
+; X86-LABEL: test_maskz_expand_load_b_512:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovq {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf8,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpexpandb (%eax), %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xc9,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_maskz_expand_load_b_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovq %rsi, %k1 # encoding: [0xc4,0xe1,0xfb,0x92,0xce]
+; X64-NEXT:    vpexpandb (%rdi), %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xc9,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <64 x i8> @llvm.x86.avx512.mask.expand.load.b.512(i8* %addr, <64 x i8> zeroinitializer, i64 %mask)
+  ret <64 x i8> %res
+}
+
+declare <64 x i8> @llvm.x86.avx512.mask.expand.load.b.512(i8* %addr, <64 x i8> %data, i64 %mask)
+
+define <64 x i8> @test_expand_load_b_512(i8* %addr, <64 x i8> %data) {
+; X86-LABEL: test_expand_load_b_512:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kxnorq %k0, %k0, %k1 # encoding: [0xc4,0xe1,0xfc,0x46,0xc8]
+; X86-NEXT:    vpexpandb (%eax), %zmm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_expand_load_b_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kxnorq %k0, %k0, %k1 # encoding: [0xc4,0xe1,0xfc,0x46,0xc8]
+; X64-NEXT:    vpexpandb (%rdi), %zmm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <64 x i8> @llvm.x86.avx512.mask.expand.load.b.512(i8* %addr, <64 x i8> %data, i64 -1)
+  ret <64 x i8> %res
+}
+
+define void @test_mask_compress_store_w_512(i8* %addr, <32 x i16> %data, i32 %mask) {
+; X86-LABEL: test_mask_compress_store_w_512:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpcompressw %zmm0, (%eax) {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x63,0x00]
+; X86-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_compress_store_w_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpcompressw %zmm0, (%rdi) {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x63,0x07]
+; X64-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X64-NEXT:    retq # encoding: [0xc3]
+  call void @llvm.x86.avx512.mask.compress.store.w.512(i8* %addr, <32 x i16> %data, i32 %mask)
+  ret void
+}
+
+declare void @llvm.x86.avx512.mask.compress.store.w.512(i8* %addr, <32 x i16> %data, i32 %mask)
+
+define <32 x i16> @test_mask_compress_w_512(<32 x i16> %data, <32 x i16> %passthru, i32 %mask) {
+; X86-LABEL: test_mask_compress_w_512:
+; X86:       # %bb.0:
+; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x04]
+; X86-NEXT:    vpcompressw %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x63,0xc1]
+; X86-NEXT:    vmovdqa64 %zmm1, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc1]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_compress_w_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
+; X64-NEXT:    vpcompressw %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x63,0xc1]
+; X64-NEXT:    vmovdqa64 %zmm1, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc1]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <32 x i16> @llvm.x86.avx512.mask.compress.w.512(<32 x i16> %data, <32 x i16> %passthru, i32 %mask)
+  ret <32 x i16> %res
+}
+
+define <32 x i16> @test_maskz_compress_w_512(<32 x i16> %data, i32 %mask) {
+; X86-LABEL: test_maskz_compress_w_512:
+; X86:       # %bb.0:
+; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x04]
+; X86-NEXT:    vpcompressw %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xc9,0x63,0xc0]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_maskz_compress_w_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %edi, %k1 # encoding: [0xc5,0xfb,0x92,0xcf]
+; X64-NEXT:    vpcompressw %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xc9,0x63,0xc0]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <32 x i16> @llvm.x86.avx512.mask.compress.w.512(<32 x i16> %data, <32 x i16> zeroinitializer, i32 %mask)
+  ret <32 x i16> %res
+}
+
+define <32 x i16> @test_compress_w_512(<32 x i16> %data) {
+; CHECK-LABEL: test_compress_w_512:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
+  %res = call <32 x i16> @llvm.x86.avx512.mask.compress.w.512(<32 x i16> %data, <32 x i16> undef, i32 -1)
+  ret <32 x i16> %res
+}
+
+declare <32 x i16> @llvm.x86.avx512.mask.compress.w.512(<32 x i16> %data, <32 x i16> %src0, i32 %mask)
+
+define void @test_compress_store_w_512(i8* %addr, <32 x i16> %data) {
+; X86-LABEL: test_compress_store_w_512:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kxnord %k0, %k0, %k1 # encoding: [0xc4,0xe1,0xfd,0x46,0xc8]
+; X86-NEXT:    vpcompressw %zmm0, (%eax) {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x63,0x00]
+; X86-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_compress_store_w_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kxnord %k0, %k0, %k1 # encoding: [0xc4,0xe1,0xfd,0x46,0xc8]
+; X64-NEXT:    vpcompressw %zmm0, (%rdi) {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x63,0x07]
+; X64-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X64-NEXT:    retq # encoding: [0xc3]
+  call void @llvm.x86.avx512.mask.compress.store.w.512(i8* %addr, <32 x i16> %data, i32 -1)
+  ret void
+}
+
+define void @test_mask_compress_store_b_512(i8* %addr, <64 x i8> %data, i64 %mask) {
+; X86-LABEL: test_mask_compress_store_b_512:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovq {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf8,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpcompressb %zmm0, (%eax) {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x63,0x00]
+; X86-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_compress_store_b_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovq %rsi, %k1 # encoding: [0xc4,0xe1,0xfb,0x92,0xce]
+; X64-NEXT:    vpcompressb %zmm0, (%rdi) {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x63,0x07]
+; X64-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X64-NEXT:    retq # encoding: [0xc3]
+  call void @llvm.x86.avx512.mask.compress.store.b.512(i8* %addr, <64 x i8> %data, i64 %mask)
+  ret void
+}
+
+declare void @llvm.x86.avx512.mask.compress.store.b.512(i8* %addr, <64 x i8> %data, i64 %mask)
+
+define <64 x i8> @test_mask_compress_b_512(<64 x i8> %data, <64 x i8> %passthru, i64 %mask) {
+; X86-LABEL: test_mask_compress_b_512:
+; X86:       # %bb.0:
+; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k0 # encoding: [0xc4,0xe1,0xf9,0x90,0x44,0x24,0x04]
+; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    kunpckdq %k0, %k1, %k1 # encoding: [0xc4,0xe1,0xf4,0x4b,0xc8]
+; X86-NEXT:    vpcompressb %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x63,0xc1]
+; X86-NEXT:    vmovdqa64 %zmm1, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc1]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_compress_b_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovq %rdi, %k1 # encoding: [0xc4,0xe1,0xfb,0x92,0xcf]
+; X64-NEXT:    vpcompressb %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x63,0xc1]
+; X64-NEXT:    vmovdqa64 %zmm1, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc1]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <64 x i8> @llvm.x86.avx512.mask.compress.b.512(<64 x i8> %data, <64 x i8> %passthru, i64 %mask)
+  ret <64 x i8> %res
+}
+
+define <64 x i8> @test_maskz_compress_b_512(<64 x i8> %data, i64 %mask) {
+; X86-LABEL: test_maskz_compress_b_512:
+; X86:       # %bb.0:
+; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k0 # encoding: [0xc4,0xe1,0xf9,0x90,0x44,0x24,0x04]
+; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    kunpckdq %k0, %k1, %k1 # encoding: [0xc4,0xe1,0xf4,0x4b,0xc8]
+; X86-NEXT:    vpcompressb %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xc9,0x63,0xc0]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_maskz_compress_b_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovq %rdi, %k1 # encoding: [0xc4,0xe1,0xfb,0x92,0xcf]
+; X64-NEXT:    vpcompressb %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xc9,0x63,0xc0]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <64 x i8> @llvm.x86.avx512.mask.compress.b.512(<64 x i8> %data, <64 x i8> zeroinitializer, i64 %mask)
+  ret <64 x i8> %res
+}
+
+define <64 x i8> @test_compress_b_512(<64 x i8> %data) {
+; CHECK-LABEL: test_compress_b_512:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
+  %res = call <64 x i8> @llvm.x86.avx512.mask.compress.b.512(<64 x i8> %data, <64 x i8> undef, i64 -1)
+  ret <64 x i8> %res
+}
+
+declare <64 x i8> @llvm.x86.avx512.mask.compress.b.512(<64 x i8> %data, <64 x i8> %src0, i64 %mask)
+
+define void @test_compress_store_b_512(i8* %addr, <64 x i8> %data) {
+; X86-LABEL: test_compress_store_b_512:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kxnorq %k0, %k0, %k1 # encoding: [0xc4,0xe1,0xfc,0x46,0xc8]
+; X86-NEXT:    vpcompressb %zmm0, (%eax) {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x63,0x00]
+; X86-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_compress_store_b_512:
+; X64:       # %bb.0:
+; X64-NEXT:    kxnorq %k0, %k0, %k1 # encoding: [0xc4,0xe1,0xfc,0x46,0xc8]
+; X64-NEXT:    vpcompressb %zmm0, (%rdi) {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x63,0x07]
+; X64-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X64-NEXT:    retq # encoding: [0xc3]
+  call void @llvm.x86.avx512.mask.compress.store.b.512(i8* %addr, <64 x i8> %data, i64 -1)
+  ret void
+}
diff --git a/llvm/test/CodeGen/X86/avx512vbmi2vl-intrinsics-upgrade.ll b/llvm/test/CodeGen/X86/avx512vbmi2vl-intrinsics-upgrade.ll
new file mode 100644
index 00000000000..d26f1648e11
--- /dev/null
+++ b/llvm/test/CodeGen/X86/avx512vbmi2vl-intrinsics-upgrade.ll
@@ -0,0 +1,370 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512vbmi2,+avx512vl --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X86
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vbmi2,+avx512vl --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X64
+
+define <8 x i16> @test_mask_expand_load_w_128(i8* %addr, <8 x i16> %data, i8 %mask) {
+; X86-LABEL: test_mask_expand_load_w_128:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx # encoding: [0x0f,0xb6,0x4c,0x24,0x08]
+; X86-NEXT:    kmovd %ecx, %k1 # encoding: [0xc5,0xfb,0x92,0xc9]
+; X86-NEXT:    vpexpandw (%eax), %xmm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_expand_load_w_128:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpexpandw (%rdi), %xmm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <8 x i16> @llvm.x86.avx512.mask.expand.load.w.128(i8* %addr, <8 x i16> %data, i8 %mask)
+  ret <8 x i16> %res
+}
+
+define <8 x i16> @test_maskz_expand_load_w_128(i8* %addr, i8 %mask) {
+; X86-LABEL: test_maskz_expand_load_w_128:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx # encoding: [0x0f,0xb6,0x4c,0x24,0x08]
+; X86-NEXT:    kmovd %ecx, %k1 # encoding: [0xc5,0xfb,0x92,0xc9]
+; X86-NEXT:    vpexpandw (%eax), %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0x89,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_maskz_expand_load_w_128:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpexpandw (%rdi), %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0x89,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <8 x i16> @llvm.x86.avx512.mask.expand.load.w.128(i8* %addr, <8 x i16> zeroinitializer, i8 %mask)
+  ret <8 x i16> %res
+}
+
+declare <8 x i16> @llvm.x86.avx512.mask.expand.load.w.128(i8* %addr, <8 x i16> %data, i8 %mask)
+
+define <8 x i16> @test_expand_load_w_128(i8* %addr, <8 x i16> %data) {
+; X86-LABEL: test_expand_load_w_128:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kxnorw %k0, %k0, %k1 # encoding: [0xc5,0xfc,0x46,0xc8]
+; X86-NEXT:    vpexpandw (%eax), %xmm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_expand_load_w_128:
+; X64:       # %bb.0:
+; X64-NEXT:    kxnorw %k0, %k0, %k1 # encoding: [0xc5,0xfc,0x46,0xc8]
+; X64-NEXT:    vpexpandw (%rdi), %xmm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <8 x i16> @llvm.x86.avx512.mask.expand.load.w.128(i8* %addr, <8 x i16> %data, i8 -1)
+  ret <8 x i16> %res
+}
+
+define <16 x i8> @test_mask_expand_load_b_128(i8* %addr, <16 x i8> %data, i16 %mask) {
+; X86-LABEL: test_mask_expand_load_b_128:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpexpandb (%eax), %xmm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_expand_load_b_128:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpexpandb (%rdi), %xmm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <16 x i8> @llvm.x86.avx512.mask.expand.load.b.128(i8* %addr, <16 x i8> %data, i16 %mask)
+  ret <16 x i8> %res
+}
+
+define <16 x i8> @test_maskz_expand_load_b_128(i8* %addr, i16 %mask) {
+; X86-LABEL: test_maskz_expand_load_b_128:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpexpandb (%eax), %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0x89,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_maskz_expand_load_b_128:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpexpandb (%rdi), %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0x89,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <16 x i8> @llvm.x86.avx512.mask.expand.load.b.128(i8* %addr, <16 x i8> zeroinitializer, i16 %mask)
+  ret <16 x i8> %res
+}
+
+declare <16 x i8> @llvm.x86.avx512.mask.expand.load.b.128(i8* %addr, <16 x i8> %data, i16 %mask)
+
+define <16 x i8> @test_expand_load_b_128(i8* %addr, <16 x i8> %data) {
+; X86-LABEL: test_expand_load_b_128:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kxnorw %k0, %k0, %k1 # encoding: [0xc5,0xfc,0x46,0xc8]
+; X86-NEXT:    vpexpandb (%eax), %xmm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_expand_load_b_128:
+; X64:       # %bb.0:
+; X64-NEXT:    kxnorw %k0, %k0, %k1 # encoding: [0xc5,0xfc,0x46,0xc8]
+; X64-NEXT:    vpexpandb (%rdi), %xmm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <16 x i8> @llvm.x86.avx512.mask.expand.load.b.128(i8* %addr, <16 x i8> %data, i16 -1)
+  ret <16 x i8> %res
+}
+
+define void @test_mask_compress_store_w_128(i8* %addr, <8 x i16> %data, i8 %mask) {
+; X86-LABEL: test_mask_compress_store_w_128:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx # encoding: [0x0f,0xb6,0x4c,0x24,0x08]
+; X86-NEXT:    kmovd %ecx, %k1 # encoding: [0xc5,0xfb,0x92,0xc9]
+; X86-NEXT:    vpcompressw %xmm0, (%eax) {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0x63,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_compress_store_w_128:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpcompressw %xmm0, (%rdi) {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0x63,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  call void @llvm.x86.avx512.mask.compress.store.w.128(i8* %addr, <8 x i16> %data, i8 %mask)
+  ret void
+}
+
+declare void @llvm.x86.avx512.mask.compress.store.w.128(i8* %addr, <8 x i16> %data, i8 %mask)
+
+define void @test_compress_store_w_128(i8* %addr, <8 x i16> %data) {
+; X86-LABEL: test_compress_store_w_128:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kxnorw %k0, %k0, %k1 # encoding: [0xc5,0xfc,0x46,0xc8]
+; X86-NEXT:    vpcompressw %xmm0, (%eax) {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0x63,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_compress_store_w_128:
+; X64:       # %bb.0:
+; X64-NEXT:    kxnorw %k0, %k0, %k1 # encoding: [0xc5,0xfc,0x46,0xc8]
+; X64-NEXT:    vpcompressw %xmm0, (%rdi) {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0x63,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  call void @llvm.x86.avx512.mask.compress.store.w.128(i8* %addr, <8 x i16> %data, i8 -1)
+  ret void
+}
+
+define void @test_mask_compress_store_b_128(i8* %addr, <16 x i8> %data, i16 %mask) {
+; X86-LABEL: test_mask_compress_store_b_128:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpcompressb %xmm0, (%eax) {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0x63,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_compress_store_b_128:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpcompressb %xmm0, (%rdi) {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0x63,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  call void @llvm.x86.avx512.mask.compress.store.b.128(i8* %addr, <16 x i8> %data, i16 %mask)
+  ret void
+}
+
+declare void @llvm.x86.avx512.mask.compress.store.b.128(i8* %addr, <16 x i8> %data, i16 %mask)
+
+define void @test_compress_store_b_128(i8* %addr, <16 x i8> %data) {
+; X86-LABEL: test_compress_store_b_128:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kxnorw %k0, %k0, %k1 # encoding: [0xc5,0xfc,0x46,0xc8]
+; X86-NEXT:    vpcompressb %xmm0, (%eax) {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0x63,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_compress_store_b_128:
+; X64:       # %bb.0:
+; X64-NEXT:    kxnorw %k0, %k0, %k1 # encoding: [0xc5,0xfc,0x46,0xc8]
+; X64-NEXT:    vpcompressb %xmm0, (%rdi) {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0x63,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  call void @llvm.x86.avx512.mask.compress.store.b.128(i8* %addr, <16 x i8> %data, i16 -1)
+  ret void
+}
+
+define <16 x i16> @test_mask_expand_load_w_256(i8* %addr, <16 x i16> %data, i16 %mask) {
+; X86-LABEL: test_mask_expand_load_w_256:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpexpandw (%eax), %ymm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_expand_load_w_256:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpexpandw (%rdi), %ymm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <16 x i16> @llvm.x86.avx512.mask.expand.load.w.256(i8* %addr, <16 x i16> %data, i16 %mask)
+  ret <16 x i16> %res
+}
+
+define <16 x i16> @test_maskz_expand_load_w_256(i8* %addr, i16 %mask) {
+; X86-LABEL: test_maskz_expand_load_w_256:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpexpandw (%eax), %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xa9,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_maskz_expand_load_w_256:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpexpandw (%rdi), %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xa9,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <16 x i16> @llvm.x86.avx512.mask.expand.load.w.256(i8* %addr, <16 x i16> zeroinitializer, i16 %mask)
+  ret <16 x i16> %res
+}
+
+declare <16 x i16> @llvm.x86.avx512.mask.expand.load.w.256(i8* %addr, <16 x i16> %data, i16 %mask)
+
+define <16 x i16> @test_expand_load_w_256(i8* %addr, <16 x i16> %data) {
+; X86-LABEL: test_expand_load_w_256:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kxnorw %k0, %k0, %k1 # encoding: [0xc5,0xfc,0x46,0xc8]
+; X86-NEXT:    vpexpandw (%eax), %ymm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_expand_load_w_256:
+; X64:       # %bb.0:
+; X64-NEXT:    kxnorw %k0, %k0, %k1 # encoding: [0xc5,0xfc,0x46,0xc8]
+; X64-NEXT:    vpexpandw (%rdi), %ymm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <16 x i16> @llvm.x86.avx512.mask.expand.load.w.256(i8* %addr, <16 x i16> %data, i16 -1)
+  ret <16 x i16> %res
+}
+
+define <32 x i8> @test_mask_expand_load_b_256(i8* %addr, <32 x i8> %data, i32 %mask) {
+; X86-LABEL: test_mask_expand_load_b_256:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpexpandb (%eax), %ymm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_expand_load_b_256:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpexpandb (%rdi), %ymm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <32 x i8> @llvm.x86.avx512.mask.expand.load.b.256(i8* %addr, <32 x i8> %data, i32 %mask)
+  ret <32 x i8> %res
+}
+
+define <32 x i8> @test_maskz_expand_load_b_256(i8* %addr, i32 %mask) {
+; X86-LABEL: test_maskz_expand_load_b_256:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpexpandb (%eax), %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xa9,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_maskz_expand_load_b_256:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpexpandb (%rdi), %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x7d,0xa9,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <32 x i8> @llvm.x86.avx512.mask.expand.load.b.256(i8* %addr, <32 x i8> zeroinitializer, i32 %mask)
+  ret <32 x i8> %res
+}
+
+declare <32 x i8> @llvm.x86.avx512.mask.expand.load.b.256(i8* %addr, <32 x i8> %data, i32 %mask)
+
+define <32 x i8> @test_expand_load_b_256(i8* %addr, <32 x i8> %data) {
+; X86-LABEL: test_expand_load_b_256:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kxnord %k0, %k0, %k1 # encoding: [0xc4,0xe1,0xfd,0x46,0xc8]
+; X86-NEXT:    vpexpandb (%eax), %ymm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0x62,0x00]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_expand_load_b_256:
+; X64:       # %bb.0:
+; X64-NEXT:    kxnord %k0, %k0, %k1 # encoding: [0xc4,0xe1,0xfd,0x46,0xc8]
+; X64-NEXT:    vpexpandb (%rdi), %ymm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0x62,0x07]
+; X64-NEXT:    retq # encoding: [0xc3]
+  %res = call <32 x i8> @llvm.x86.avx512.mask.expand.load.b.256(i8* %addr, <32 x i8> %data, i32 -1)
+  ret <32 x i8> %res
+}
+
+define void @test_mask_compress_store_w_256(i8* %addr, <16 x i16> %data, i16 %mask) {
+; X86-LABEL: test_mask_compress_store_w_256:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpcompressw %ymm0, (%eax) {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x63,0x00]
+; X86-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_compress_store_w_256:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpcompressw %ymm0, (%rdi) {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x63,0x07]
+; X64-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X64-NEXT:    retq # encoding: [0xc3]
+  call void @llvm.x86.avx512.mask.compress.store.w.256(i8* %addr, <16 x i16> %data, i16 %mask)
+  ret void
+}
+
+declare void @llvm.x86.avx512.mask.compress.store.w.256(i8* %addr, <16 x i16> %data, i16 %mask)
+
+define void @test_compress_store_w_256(i8* %addr, <16 x i16> %data) {
+; X86-LABEL: test_compress_store_w_256:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kxnorw %k0, %k0, %k1 # encoding: [0xc5,0xfc,0x46,0xc8]
+; X86-NEXT:    vpcompressw %ymm0, (%eax) {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x63,0x00]
+; X86-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_compress_store_w_256:
+; X64:       # %bb.0:
+; X64-NEXT:    kxnorw %k0, %k0, %k1 # encoding: [0xc5,0xfc,0x46,0xc8]
+; X64-NEXT:    vpcompressw %ymm0, (%rdi) {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x63,0x07]
+; X64-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X64-NEXT:    retq # encoding: [0xc3]
+  call void @llvm.x86.avx512.mask.compress.store.w.256(i8* %addr, <16 x i16> %data, i16 -1)
+  ret void
+}
+
+define void @test_mask_compress_store_b_256(i8* %addr, <32 x i8> %data, i32 %mask) {
+; X86-LABEL: test_mask_compress_store_b_256:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kmovd {{[0-9]+}}(%esp), %k1 # encoding: [0xc4,0xe1,0xf9,0x90,0x4c,0x24,0x08]
+; X86-NEXT:    vpcompressb %ymm0, (%eax) {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0x63,0x00]
+; X86-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_mask_compress_store_b_256:
+; X64:       # %bb.0:
+; X64-NEXT:    kmovd %esi, %k1 # encoding: [0xc5,0xfb,0x92,0xce]
+; X64-NEXT:    vpcompressb %ymm0, (%rdi) {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0x63,0x07]
+; X64-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X64-NEXT:    retq # encoding: [0xc3]
+  call void @llvm.x86.avx512.mask.compress.store.b.256(i8* %addr, <32 x i8> %data, i32 %mask)
+  ret void
+}
+
+declare void @llvm.x86.avx512.mask.compress.store.b.256(i8* %addr, <32 x i8> %data, i32 %mask)
+
+define void @test_compress_store_b_256(i8* %addr, <32 x i8> %data) {
+; X86-LABEL: test_compress_store_b_256:
+; X86:       # %bb.0:
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
+; X86-NEXT:    kxnord %k0, %k0, %k1 # encoding: [0xc4,0xe1,0xfd,0x46,0xc8]
+; X86-NEXT:    vpcompressb %ymm0, (%eax) {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0x63,0x00]
+; X86-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X86-NEXT:    retl # encoding: [0xc3]
+;
+; X64-LABEL: test_compress_store_b_256:
+; X64:       # %bb.0:
+; X64-NEXT:    kxnord %k0, %k0, %k1 # encoding: [0xc4,0xe1,0xfd,0x46,0xc8]
+; X64-NEXT:    vpcompressb %ymm0, (%rdi) {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0x63,0x07]
+; X64-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
+; X64-NEXT:    retq # encoding: [0xc3]
+  call void @llvm.x86.avx512.mask.compress.store.b.256(i8* %addr, <32 x i8> %data, i32 -1)
+  ret void
+}