[llvm] 48a0bb7 - [X86] masked_store_trunc.ll - add missing prefixes (#160685)
via llvm-commits
llvm-commits at lists.llvm.org
Thu Sep 25 06:18:51 PDT 2025
Author: Simon Pilgrim
Date: 2025-09-25T13:18:47Z
New Revision: 48a0bb71053bf4dd7884cae998ed341df02ecef3
URL: https://github.com/llvm/llvm-project/commit/48a0bb71053bf4dd7884cae998ed341df02ecef3
DIFF: https://github.com/llvm/llvm-project/commit/48a0bb71053bf4dd7884cae998ed341df02ecef3.diff
LOG: [X86] masked_store_trunc.ll - add missing prefixes (#160685)
Since #159321 we now get actual warnings when we're missing coverage
Added:
Modified:
llvm/test/CodeGen/X86/masked_store_trunc.ll
Removed:
################################################################################
diff --git a/llvm/test/CodeGen/X86/masked_store_trunc.ll b/llvm/test/CodeGen/X86/masked_store_trunc.ll
index 2f0d419132492..ecf4fbb603a8f 100644
--- a/llvm/test/CodeGen/X86/masked_store_trunc.ll
+++ b/llvm/test/CodeGen/X86/masked_store_trunc.ll
@@ -1,11 +1,11 @@
; 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=sse4.2 | FileCheck %s --check-prefix=SSE4
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=sse2 | FileCheck %s --check-prefixes=SSE2
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=sse4.2 | FileCheck %s --check-prefixes=SSE4
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx | FileCheck %s --check-prefixes=AVX,AVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx2 | FileCheck %s --check-prefixes=AVX,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=avx512vl | FileCheck %s --check-prefix=AVX512VL
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512bw | FileCheck %s --check-prefix=AVX512BW
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512f | FileCheck %s --check-prefixes=AVX512,AVX512F
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512vl | FileCheck %s --check-prefixes=AVX512VL,AVX512FVL
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512bw | FileCheck %s --check-prefixes=AVX512,AVX512BW
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512vl,avx512bw | FileCheck %s --check-prefixes=AVX512VL,AVX512BWVL
define void @truncstore_v8i64_v8i32(<8 x i64> %x, ptr %p, <8 x i32> %mask) {
@@ -178,13 +178,13 @@ define void @truncstore_v8i64_v8i32(<8 x i64> %x, ptr %p, <8 x i32> %mask) {
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
-; AVX512F-LABEL: truncstore_v8i64_v8i32:
-; AVX512F: # %bb.0:
-; AVX512F-NEXT: # kill: def $ymm1 killed $ymm1 def $zmm1
-; AVX512F-NEXT: vptestmd %zmm1, %zmm1, %k1
-; AVX512F-NEXT: vpmovqd %zmm0, (%rdi) {%k1}
-; AVX512F-NEXT: vzeroupper
-; AVX512F-NEXT: retq
+; AVX512-LABEL: truncstore_v8i64_v8i32:
+; AVX512: # %bb.0:
+; AVX512-NEXT: # kill: def $ymm1 killed $ymm1 def $zmm1
+; AVX512-NEXT: vptestmd %zmm1, %zmm1, %k1
+; AVX512-NEXT: vpmovqd %zmm0, (%rdi) {%k1}
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
;
; AVX512VL-LABEL: truncstore_v8i64_v8i32:
; AVX512VL: # %bb.0:
@@ -192,14 +192,6 @@ define void @truncstore_v8i64_v8i32(<8 x i64> %x, ptr %p, <8 x i32> %mask) {
; AVX512VL-NEXT: vpmovqd %zmm0, (%rdi) {%k1}
; AVX512VL-NEXT: vzeroupper
; AVX512VL-NEXT: retq
-;
-; AVX512BW-LABEL: truncstore_v8i64_v8i32:
-; AVX512BW: # %bb.0:
-; AVX512BW-NEXT: # kill: def $ymm1 killed $ymm1 def $zmm1
-; AVX512BW-NEXT: vptestmd %zmm1, %zmm1, %k1
-; AVX512BW-NEXT: vpmovqd %zmm0, (%rdi) {%k1}
-; AVX512BW-NEXT: vzeroupper
-; AVX512BW-NEXT: retq
%a = icmp ne <8 x i32> %mask, zeroinitializer
%b = trunc <8 x i64> %x to <8 x i32>
call void @llvm.masked.store.v8i32.p0(<8 x i32> %b, ptr %p, i32 1, <8 x i1> %a)
@@ -573,6 +565,70 @@ define void @truncstore_v8i64_v8i16(<8 x i64> %x, ptr %p, <8 x i32> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v8i64_v8i16:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vptestmd %ymm1, %ymm1, %k0
+; AVX512FVL-NEXT: vpmovqw %zmm0, %xmm0
+; AVX512FVL-NEXT: kmovw %k0, %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB1_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB1_3
+; AVX512FVL-NEXT: .LBB1_4: # %else2
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: jne .LBB1_5
+; AVX512FVL-NEXT: .LBB1_6: # %else4
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: jne .LBB1_7
+; AVX512FVL-NEXT: .LBB1_8: # %else6
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: jne .LBB1_9
+; AVX512FVL-NEXT: .LBB1_10: # %else8
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: jne .LBB1_11
+; AVX512FVL-NEXT: .LBB1_12: # %else10
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: jne .LBB1_13
+; AVX512FVL-NEXT: .LBB1_14: # %else12
+; AVX512FVL-NEXT: testb $-128, %al
+; AVX512FVL-NEXT: jne .LBB1_15
+; AVX512FVL-NEXT: .LBB1_16: # %else14
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB1_1: # %cond.store
+; AVX512FVL-NEXT: vpextrw $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB1_4
+; AVX512FVL-NEXT: .LBB1_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrw $1, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: je .LBB1_6
+; AVX512FVL-NEXT: .LBB1_5: # %cond.store3
+; AVX512FVL-NEXT: vpextrw $2, %xmm0, 4(%rdi)
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: je .LBB1_8
+; AVX512FVL-NEXT: .LBB1_7: # %cond.store5
+; AVX512FVL-NEXT: vpextrw $3, %xmm0, 6(%rdi)
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: je .LBB1_10
+; AVX512FVL-NEXT: .LBB1_9: # %cond.store7
+; AVX512FVL-NEXT: vpextrw $4, %xmm0, 8(%rdi)
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: je .LBB1_12
+; AVX512FVL-NEXT: .LBB1_11: # %cond.store9
+; AVX512FVL-NEXT: vpextrw $5, %xmm0, 10(%rdi)
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: je .LBB1_14
+; AVX512FVL-NEXT: .LBB1_13: # %cond.store11
+; AVX512FVL-NEXT: vpextrw $6, %xmm0, 12(%rdi)
+; AVX512FVL-NEXT: testb $-128, %al
+; AVX512FVL-NEXT: je .LBB1_16
+; AVX512FVL-NEXT: .LBB1_15: # %cond.store13
+; AVX512FVL-NEXT: vpextrw $7, %xmm0, 14(%rdi)
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v8i64_v8i16:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $ymm1 killed $ymm1 def $zmm1
@@ -960,6 +1016,70 @@ define void @truncstore_v8i64_v8i8(<8 x i64> %x, ptr %p, <8 x i32> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v8i64_v8i8:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vptestmd %ymm1, %ymm1, %k0
+; AVX512FVL-NEXT: vpmovqb %zmm0, %xmm0
+; AVX512FVL-NEXT: kmovw %k0, %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB2_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB2_3
+; AVX512FVL-NEXT: .LBB2_4: # %else2
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: jne .LBB2_5
+; AVX512FVL-NEXT: .LBB2_6: # %else4
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: jne .LBB2_7
+; AVX512FVL-NEXT: .LBB2_8: # %else6
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: jne .LBB2_9
+; AVX512FVL-NEXT: .LBB2_10: # %else8
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: jne .LBB2_11
+; AVX512FVL-NEXT: .LBB2_12: # %else10
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: jne .LBB2_13
+; AVX512FVL-NEXT: .LBB2_14: # %else12
+; AVX512FVL-NEXT: testb $-128, %al
+; AVX512FVL-NEXT: jne .LBB2_15
+; AVX512FVL-NEXT: .LBB2_16: # %else14
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB2_1: # %cond.store
+; AVX512FVL-NEXT: vpextrb $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB2_4
+; AVX512FVL-NEXT: .LBB2_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrb $1, %xmm0, 1(%rdi)
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: je .LBB2_6
+; AVX512FVL-NEXT: .LBB2_5: # %cond.store3
+; AVX512FVL-NEXT: vpextrb $2, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: je .LBB2_8
+; AVX512FVL-NEXT: .LBB2_7: # %cond.store5
+; AVX512FVL-NEXT: vpextrb $3, %xmm0, 3(%rdi)
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: je .LBB2_10
+; AVX512FVL-NEXT: .LBB2_9: # %cond.store7
+; AVX512FVL-NEXT: vpextrb $4, %xmm0, 4(%rdi)
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: je .LBB2_12
+; AVX512FVL-NEXT: .LBB2_11: # %cond.store9
+; AVX512FVL-NEXT: vpextrb $5, %xmm0, 5(%rdi)
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: je .LBB2_14
+; AVX512FVL-NEXT: .LBB2_13: # %cond.store11
+; AVX512FVL-NEXT: vpextrb $6, %xmm0, 6(%rdi)
+; AVX512FVL-NEXT: testb $-128, %al
+; AVX512FVL-NEXT: je .LBB2_16
+; AVX512FVL-NEXT: .LBB2_15: # %cond.store13
+; AVX512FVL-NEXT: vpextrb $7, %xmm0, 7(%rdi)
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v8i64_v8i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $ymm1 killed $ymm1 def $zmm1
@@ -1080,17 +1200,17 @@ define void @truncstore_v4i64_v4i32(<4 x i64> %x, ptr %p, <4 x i32> %mask) {
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
-; AVX512F-LABEL: truncstore_v4i64_v4i32:
-; AVX512F: # %bb.0:
-; AVX512F-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
-; AVX512F-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0
-; AVX512F-NEXT: vptestmd %zmm1, %zmm1, %k0
-; AVX512F-NEXT: kshiftlw $12, %k0, %k0
-; AVX512F-NEXT: kshiftrw $12, %k0, %k1
-; AVX512F-NEXT: vpmovqd %zmm0, %ymm0
-; AVX512F-NEXT: vmovdqu32 %zmm0, (%rdi) {%k1}
-; AVX512F-NEXT: vzeroupper
-; AVX512F-NEXT: retq
+; AVX512-LABEL: truncstore_v4i64_v4i32:
+; AVX512: # %bb.0:
+; AVX512-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
+; AVX512-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0
+; AVX512-NEXT: vptestmd %zmm1, %zmm1, %k0
+; AVX512-NEXT: kshiftlw $12, %k0, %k0
+; AVX512-NEXT: kshiftrw $12, %k0, %k1
+; AVX512-NEXT: vpmovqd %zmm0, %ymm0
+; AVX512-NEXT: vmovdqu32 %zmm0, (%rdi) {%k1}
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
;
; AVX512VL-LABEL: truncstore_v4i64_v4i32:
; AVX512VL: # %bb.0:
@@ -1098,18 +1218,6 @@ define void @truncstore_v4i64_v4i32(<4 x i64> %x, ptr %p, <4 x i32> %mask) {
; AVX512VL-NEXT: vpmovqd %ymm0, (%rdi) {%k1}
; AVX512VL-NEXT: vzeroupper
; AVX512VL-NEXT: retq
-;
-; AVX512BW-LABEL: truncstore_v4i64_v4i32:
-; AVX512BW: # %bb.0:
-; AVX512BW-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
-; AVX512BW-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0
-; AVX512BW-NEXT: vptestmd %zmm1, %zmm1, %k0
-; AVX512BW-NEXT: kshiftlw $12, %k0, %k0
-; AVX512BW-NEXT: kshiftrw $12, %k0, %k1
-; AVX512BW-NEXT: vpmovqd %zmm0, %ymm0
-; AVX512BW-NEXT: vmovdqu32 %zmm0, (%rdi) {%k1}
-; AVX512BW-NEXT: vzeroupper
-; AVX512BW-NEXT: retq
%a = icmp ne <4 x i32> %mask, zeroinitializer
%b = trunc <4 x i64> %x to <4 x i32>
call void @llvm.masked.store.v4i32.p0(<4 x i32> %b, ptr %p, i32 1, <4 x i1> %a)
@@ -1321,6 +1429,42 @@ define void @truncstore_v4i64_v4i16(<4 x i64> %x, ptr %p, <4 x i32> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v4i64_v4i16:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vptestmd %xmm1, %xmm1, %k0
+; AVX512FVL-NEXT: vpmovqw %ymm0, %xmm0
+; AVX512FVL-NEXT: kmovw %k0, %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB4_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB4_3
+; AVX512FVL-NEXT: .LBB4_4: # %else2
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: jne .LBB4_5
+; AVX512FVL-NEXT: .LBB4_6: # %else4
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: jne .LBB4_7
+; AVX512FVL-NEXT: .LBB4_8: # %else6
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB4_1: # %cond.store
+; AVX512FVL-NEXT: vpextrw $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB4_4
+; AVX512FVL-NEXT: .LBB4_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrw $1, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: je .LBB4_6
+; AVX512FVL-NEXT: .LBB4_5: # %cond.store3
+; AVX512FVL-NEXT: vpextrw $2, %xmm0, 4(%rdi)
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: je .LBB4_8
+; AVX512FVL-NEXT: .LBB4_7: # %cond.store5
+; AVX512FVL-NEXT: vpextrw $3, %xmm0, 6(%rdi)
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v4i64_v4i16:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
@@ -1552,6 +1696,42 @@ define void @truncstore_v4i64_v4i8(<4 x i64> %x, ptr %p, <4 x i32> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v4i64_v4i8:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vptestmd %xmm1, %xmm1, %k0
+; AVX512FVL-NEXT: vpmovqb %ymm0, %xmm0
+; AVX512FVL-NEXT: kmovw %k0, %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB5_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB5_3
+; AVX512FVL-NEXT: .LBB5_4: # %else2
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: jne .LBB5_5
+; AVX512FVL-NEXT: .LBB5_6: # %else4
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: jne .LBB5_7
+; AVX512FVL-NEXT: .LBB5_8: # %else6
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB5_1: # %cond.store
+; AVX512FVL-NEXT: vpextrb $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB5_4
+; AVX512FVL-NEXT: .LBB5_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrb $1, %xmm0, 1(%rdi)
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: je .LBB5_6
+; AVX512FVL-NEXT: .LBB5_5: # %cond.store3
+; AVX512FVL-NEXT: vpextrb $2, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: je .LBB5_8
+; AVX512FVL-NEXT: .LBB5_7: # %cond.store5
+; AVX512FVL-NEXT: vpextrb $3, %xmm0, 3(%rdi)
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v4i64_v4i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
@@ -1646,33 +1826,22 @@ define void @truncstore_v2i64_v2i32(<2 x i64> %x, ptr %p, <2 x i64> %mask) {
; AVX2-NEXT: vpmaskmovd %xmm0, %xmm1, (%rdi)
; AVX2-NEXT: retq
;
-; AVX512F-LABEL: truncstore_v2i64_v2i32:
-; AVX512F: # %bb.0:
-; AVX512F-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
-; AVX512F-NEXT: vptestmq %zmm1, %zmm1, %k0
-; AVX512F-NEXT: kshiftlw $14, %k0, %k0
-; AVX512F-NEXT: kshiftrw $14, %k0, %k1
-; AVX512F-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; AVX512F-NEXT: vmovdqu32 %zmm0, (%rdi) {%k1}
-; AVX512F-NEXT: vzeroupper
-; AVX512F-NEXT: retq
+; AVX512-LABEL: truncstore_v2i64_v2i32:
+; AVX512: # %bb.0:
+; AVX512-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
+; AVX512-NEXT: vptestmq %zmm1, %zmm1, %k0
+; AVX512-NEXT: kshiftlw $14, %k0, %k0
+; AVX512-NEXT: kshiftrw $14, %k0, %k1
+; AVX512-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512-NEXT: vmovdqu32 %zmm0, (%rdi) {%k1}
+; AVX512-NEXT: vzeroupper
+; AVX512-NEXT: retq
;
; AVX512VL-LABEL: truncstore_v2i64_v2i32:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vptestmq %xmm1, %xmm1, %k1
; AVX512VL-NEXT: vpmovqd %xmm0, (%rdi) {%k1}
; AVX512VL-NEXT: retq
-;
-; AVX512BW-LABEL: truncstore_v2i64_v2i32:
-; AVX512BW: # %bb.0:
-; AVX512BW-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
-; AVX512BW-NEXT: vptestmq %zmm1, %zmm1, %k0
-; AVX512BW-NEXT: kshiftlw $14, %k0, %k0
-; AVX512BW-NEXT: kshiftrw $14, %k0, %k1
-; AVX512BW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; AVX512BW-NEXT: vmovdqu32 %zmm0, (%rdi) {%k1}
-; AVX512BW-NEXT: vzeroupper
-; AVX512BW-NEXT: retq
%a = icmp ne <2 x i64> %mask, zeroinitializer
%b = trunc <2 x i64> %x to <2 x i32>
call void @llvm.masked.store.v2i32.p0(<2 x i32> %b, ptr %p, i32 1, <2 x i1> %a)
@@ -1777,6 +1946,26 @@ define void @truncstore_v2i64_v2i16(<2 x i64> %x, ptr %p, <2 x i64> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v2i64_v2i16:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vptestmq %xmm1, %xmm1, %k0
+; AVX512FVL-NEXT: vpmovqw %xmm0, %xmm0
+; AVX512FVL-NEXT: kmovw %k0, %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB7_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB7_3
+; AVX512FVL-NEXT: .LBB7_4: # %else2
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB7_1: # %cond.store
+; AVX512FVL-NEXT: vpextrw $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB7_4
+; AVX512FVL-NEXT: .LBB7_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrw $1, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v2i64_v2i16:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
@@ -1896,6 +2085,26 @@ define void @truncstore_v2i64_v2i8(<2 x i64> %x, ptr %p, <2 x i64> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v2i64_v2i8:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vptestmq %xmm1, %xmm1, %k0
+; AVX512FVL-NEXT: vpmovqb %xmm0, %xmm0
+; AVX512FVL-NEXT: kmovw %k0, %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB8_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB8_3
+; AVX512FVL-NEXT: .LBB8_4: # %else2
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB8_1: # %cond.store
+; AVX512FVL-NEXT: vpextrb $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB8_4
+; AVX512FVL-NEXT: .LBB8_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrb $1, %xmm0, 1(%rdi)
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v2i64_v2i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
@@ -2581,6 +2790,126 @@ define void @truncstore_v16i32_v16i16(<16 x i32> %x, ptr %p, <16 x i32> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v16i32_v16i16:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vptestmd %zmm1, %zmm1, %k0
+; AVX512FVL-NEXT: vpmovdw %zmm0, %ymm0
+; AVX512FVL-NEXT: kmovw %k0, %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB9_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB9_3
+; AVX512FVL-NEXT: .LBB9_4: # %else2
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: jne .LBB9_5
+; AVX512FVL-NEXT: .LBB9_6: # %else4
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: jne .LBB9_7
+; AVX512FVL-NEXT: .LBB9_8: # %else6
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: jne .LBB9_9
+; AVX512FVL-NEXT: .LBB9_10: # %else8
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: jne .LBB9_11
+; AVX512FVL-NEXT: .LBB9_12: # %else10
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: jne .LBB9_13
+; AVX512FVL-NEXT: .LBB9_14: # %else12
+; AVX512FVL-NEXT: testb %al, %al
+; AVX512FVL-NEXT: jns .LBB9_16
+; AVX512FVL-NEXT: .LBB9_15: # %cond.store13
+; AVX512FVL-NEXT: vpextrw $7, %xmm0, 14(%rdi)
+; AVX512FVL-NEXT: .LBB9_16: # %else14
+; AVX512FVL-NEXT: testl $256, %eax # imm = 0x100
+; AVX512FVL-NEXT: vextracti128 $1, %ymm0, %xmm0
+; AVX512FVL-NEXT: jne .LBB9_17
+; AVX512FVL-NEXT: # %bb.18: # %else16
+; AVX512FVL-NEXT: testl $512, %eax # imm = 0x200
+; AVX512FVL-NEXT: jne .LBB9_19
+; AVX512FVL-NEXT: .LBB9_20: # %else18
+; AVX512FVL-NEXT: testl $1024, %eax # imm = 0x400
+; AVX512FVL-NEXT: jne .LBB9_21
+; AVX512FVL-NEXT: .LBB9_22: # %else20
+; AVX512FVL-NEXT: testl $2048, %eax # imm = 0x800
+; AVX512FVL-NEXT: jne .LBB9_23
+; AVX512FVL-NEXT: .LBB9_24: # %else22
+; AVX512FVL-NEXT: testl $4096, %eax # imm = 0x1000
+; AVX512FVL-NEXT: jne .LBB9_25
+; AVX512FVL-NEXT: .LBB9_26: # %else24
+; AVX512FVL-NEXT: testl $8192, %eax # imm = 0x2000
+; AVX512FVL-NEXT: jne .LBB9_27
+; AVX512FVL-NEXT: .LBB9_28: # %else26
+; AVX512FVL-NEXT: testl $16384, %eax # imm = 0x4000
+; AVX512FVL-NEXT: jne .LBB9_29
+; AVX512FVL-NEXT: .LBB9_30: # %else28
+; AVX512FVL-NEXT: testl $32768, %eax # imm = 0x8000
+; AVX512FVL-NEXT: jne .LBB9_31
+; AVX512FVL-NEXT: .LBB9_32: # %else30
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB9_1: # %cond.store
+; AVX512FVL-NEXT: vpextrw $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB9_4
+; AVX512FVL-NEXT: .LBB9_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrw $1, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: je .LBB9_6
+; AVX512FVL-NEXT: .LBB9_5: # %cond.store3
+; AVX512FVL-NEXT: vpextrw $2, %xmm0, 4(%rdi)
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: je .LBB9_8
+; AVX512FVL-NEXT: .LBB9_7: # %cond.store5
+; AVX512FVL-NEXT: vpextrw $3, %xmm0, 6(%rdi)
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: je .LBB9_10
+; AVX512FVL-NEXT: .LBB9_9: # %cond.store7
+; AVX512FVL-NEXT: vpextrw $4, %xmm0, 8(%rdi)
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: je .LBB9_12
+; AVX512FVL-NEXT: .LBB9_11: # %cond.store9
+; AVX512FVL-NEXT: vpextrw $5, %xmm0, 10(%rdi)
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: je .LBB9_14
+; AVX512FVL-NEXT: .LBB9_13: # %cond.store11
+; AVX512FVL-NEXT: vpextrw $6, %xmm0, 12(%rdi)
+; AVX512FVL-NEXT: testb %al, %al
+; AVX512FVL-NEXT: js .LBB9_15
+; AVX512FVL-NEXT: jmp .LBB9_16
+; AVX512FVL-NEXT: .LBB9_17: # %cond.store15
+; AVX512FVL-NEXT: vpextrw $0, %xmm0, 16(%rdi)
+; AVX512FVL-NEXT: testl $512, %eax # imm = 0x200
+; AVX512FVL-NEXT: je .LBB9_20
+; AVX512FVL-NEXT: .LBB9_19: # %cond.store17
+; AVX512FVL-NEXT: vpextrw $1, %xmm0, 18(%rdi)
+; AVX512FVL-NEXT: testl $1024, %eax # imm = 0x400
+; AVX512FVL-NEXT: je .LBB9_22
+; AVX512FVL-NEXT: .LBB9_21: # %cond.store19
+; AVX512FVL-NEXT: vpextrw $2, %xmm0, 20(%rdi)
+; AVX512FVL-NEXT: testl $2048, %eax # imm = 0x800
+; AVX512FVL-NEXT: je .LBB9_24
+; AVX512FVL-NEXT: .LBB9_23: # %cond.store21
+; AVX512FVL-NEXT: vpextrw $3, %xmm0, 22(%rdi)
+; AVX512FVL-NEXT: testl $4096, %eax # imm = 0x1000
+; AVX512FVL-NEXT: je .LBB9_26
+; AVX512FVL-NEXT: .LBB9_25: # %cond.store23
+; AVX512FVL-NEXT: vpextrw $4, %xmm0, 24(%rdi)
+; AVX512FVL-NEXT: testl $8192, %eax # imm = 0x2000
+; AVX512FVL-NEXT: je .LBB9_28
+; AVX512FVL-NEXT: .LBB9_27: # %cond.store25
+; AVX512FVL-NEXT: vpextrw $5, %xmm0, 26(%rdi)
+; AVX512FVL-NEXT: testl $16384, %eax # imm = 0x4000
+; AVX512FVL-NEXT: je .LBB9_30
+; AVX512FVL-NEXT: .LBB9_29: # %cond.store27
+; AVX512FVL-NEXT: vpextrw $6, %xmm0, 28(%rdi)
+; AVX512FVL-NEXT: testl $32768, %eax # imm = 0x8000
+; AVX512FVL-NEXT: je .LBB9_32
+; AVX512FVL-NEXT: .LBB9_31: # %cond.store29
+; AVX512FVL-NEXT: vpextrw $7, %xmm0, 30(%rdi)
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v16i32_v16i16:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vptestmd %zmm1, %zmm1, %k1
@@ -3247,6 +3576,126 @@ define void @truncstore_v16i32_v16i8(<16 x i32> %x, ptr %p, <16 x i32> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v16i32_v16i8:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vptestmd %zmm1, %zmm1, %k0
+; AVX512FVL-NEXT: vpmovdb %zmm0, %xmm0
+; AVX512FVL-NEXT: kmovw %k0, %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB10_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB10_3
+; AVX512FVL-NEXT: .LBB10_4: # %else2
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: jne .LBB10_5
+; AVX512FVL-NEXT: .LBB10_6: # %else4
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: jne .LBB10_7
+; AVX512FVL-NEXT: .LBB10_8: # %else6
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: jne .LBB10_9
+; AVX512FVL-NEXT: .LBB10_10: # %else8
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: jne .LBB10_11
+; AVX512FVL-NEXT: .LBB10_12: # %else10
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: jne .LBB10_13
+; AVX512FVL-NEXT: .LBB10_14: # %else12
+; AVX512FVL-NEXT: testb %al, %al
+; AVX512FVL-NEXT: js .LBB10_15
+; AVX512FVL-NEXT: .LBB10_16: # %else14
+; AVX512FVL-NEXT: testl $256, %eax # imm = 0x100
+; AVX512FVL-NEXT: jne .LBB10_17
+; AVX512FVL-NEXT: .LBB10_18: # %else16
+; AVX512FVL-NEXT: testl $512, %eax # imm = 0x200
+; AVX512FVL-NEXT: jne .LBB10_19
+; AVX512FVL-NEXT: .LBB10_20: # %else18
+; AVX512FVL-NEXT: testl $1024, %eax # imm = 0x400
+; AVX512FVL-NEXT: jne .LBB10_21
+; AVX512FVL-NEXT: .LBB10_22: # %else20
+; AVX512FVL-NEXT: testl $2048, %eax # imm = 0x800
+; AVX512FVL-NEXT: jne .LBB10_23
+; AVX512FVL-NEXT: .LBB10_24: # %else22
+; AVX512FVL-NEXT: testl $4096, %eax # imm = 0x1000
+; AVX512FVL-NEXT: jne .LBB10_25
+; AVX512FVL-NEXT: .LBB10_26: # %else24
+; AVX512FVL-NEXT: testl $8192, %eax # imm = 0x2000
+; AVX512FVL-NEXT: jne .LBB10_27
+; AVX512FVL-NEXT: .LBB10_28: # %else26
+; AVX512FVL-NEXT: testl $16384, %eax # imm = 0x4000
+; AVX512FVL-NEXT: jne .LBB10_29
+; AVX512FVL-NEXT: .LBB10_30: # %else28
+; AVX512FVL-NEXT: testl $32768, %eax # imm = 0x8000
+; AVX512FVL-NEXT: jne .LBB10_31
+; AVX512FVL-NEXT: .LBB10_32: # %else30
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB10_1: # %cond.store
+; AVX512FVL-NEXT: vpextrb $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB10_4
+; AVX512FVL-NEXT: .LBB10_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrb $1, %xmm0, 1(%rdi)
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: je .LBB10_6
+; AVX512FVL-NEXT: .LBB10_5: # %cond.store3
+; AVX512FVL-NEXT: vpextrb $2, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: je .LBB10_8
+; AVX512FVL-NEXT: .LBB10_7: # %cond.store5
+; AVX512FVL-NEXT: vpextrb $3, %xmm0, 3(%rdi)
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: je .LBB10_10
+; AVX512FVL-NEXT: .LBB10_9: # %cond.store7
+; AVX512FVL-NEXT: vpextrb $4, %xmm0, 4(%rdi)
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: je .LBB10_12
+; AVX512FVL-NEXT: .LBB10_11: # %cond.store9
+; AVX512FVL-NEXT: vpextrb $5, %xmm0, 5(%rdi)
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: je .LBB10_14
+; AVX512FVL-NEXT: .LBB10_13: # %cond.store11
+; AVX512FVL-NEXT: vpextrb $6, %xmm0, 6(%rdi)
+; AVX512FVL-NEXT: testb %al, %al
+; AVX512FVL-NEXT: jns .LBB10_16
+; AVX512FVL-NEXT: .LBB10_15: # %cond.store13
+; AVX512FVL-NEXT: vpextrb $7, %xmm0, 7(%rdi)
+; AVX512FVL-NEXT: testl $256, %eax # imm = 0x100
+; AVX512FVL-NEXT: je .LBB10_18
+; AVX512FVL-NEXT: .LBB10_17: # %cond.store15
+; AVX512FVL-NEXT: vpextrb $8, %xmm0, 8(%rdi)
+; AVX512FVL-NEXT: testl $512, %eax # imm = 0x200
+; AVX512FVL-NEXT: je .LBB10_20
+; AVX512FVL-NEXT: .LBB10_19: # %cond.store17
+; AVX512FVL-NEXT: vpextrb $9, %xmm0, 9(%rdi)
+; AVX512FVL-NEXT: testl $1024, %eax # imm = 0x400
+; AVX512FVL-NEXT: je .LBB10_22
+; AVX512FVL-NEXT: .LBB10_21: # %cond.store19
+; AVX512FVL-NEXT: vpextrb $10, %xmm0, 10(%rdi)
+; AVX512FVL-NEXT: testl $2048, %eax # imm = 0x800
+; AVX512FVL-NEXT: je .LBB10_24
+; AVX512FVL-NEXT: .LBB10_23: # %cond.store21
+; AVX512FVL-NEXT: vpextrb $11, %xmm0, 11(%rdi)
+; AVX512FVL-NEXT: testl $4096, %eax # imm = 0x1000
+; AVX512FVL-NEXT: je .LBB10_26
+; AVX512FVL-NEXT: .LBB10_25: # %cond.store23
+; AVX512FVL-NEXT: vpextrb $12, %xmm0, 12(%rdi)
+; AVX512FVL-NEXT: testl $8192, %eax # imm = 0x2000
+; AVX512FVL-NEXT: je .LBB10_28
+; AVX512FVL-NEXT: .LBB10_27: # %cond.store25
+; AVX512FVL-NEXT: vpextrb $13, %xmm0, 13(%rdi)
+; AVX512FVL-NEXT: testl $16384, %eax # imm = 0x4000
+; AVX512FVL-NEXT: je .LBB10_30
+; AVX512FVL-NEXT: .LBB10_29: # %cond.store27
+; AVX512FVL-NEXT: vpextrb $14, %xmm0, 14(%rdi)
+; AVX512FVL-NEXT: testl $32768, %eax # imm = 0x8000
+; AVX512FVL-NEXT: je .LBB10_32
+; AVX512FVL-NEXT: .LBB10_31: # %cond.store29
+; AVX512FVL-NEXT: vpextrb $15, %xmm0, 15(%rdi)
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v16i32_v16i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vptestmd %zmm1, %zmm1, %k1
@@ -3619,6 +4068,70 @@ define void @truncstore_v8i32_v8i16(<8 x i32> %x, ptr %p, <8 x i32> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v8i32_v8i16:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vptestmd %ymm1, %ymm1, %k0
+; AVX512FVL-NEXT: vpmovdw %ymm0, %xmm0
+; AVX512FVL-NEXT: kmovw %k0, %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB11_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB11_3
+; AVX512FVL-NEXT: .LBB11_4: # %else2
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: jne .LBB11_5
+; AVX512FVL-NEXT: .LBB11_6: # %else4
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: jne .LBB11_7
+; AVX512FVL-NEXT: .LBB11_8: # %else6
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: jne .LBB11_9
+; AVX512FVL-NEXT: .LBB11_10: # %else8
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: jne .LBB11_11
+; AVX512FVL-NEXT: .LBB11_12: # %else10
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: jne .LBB11_13
+; AVX512FVL-NEXT: .LBB11_14: # %else12
+; AVX512FVL-NEXT: testb $-128, %al
+; AVX512FVL-NEXT: jne .LBB11_15
+; AVX512FVL-NEXT: .LBB11_16: # %else14
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB11_1: # %cond.store
+; AVX512FVL-NEXT: vpextrw $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB11_4
+; AVX512FVL-NEXT: .LBB11_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrw $1, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: je .LBB11_6
+; AVX512FVL-NEXT: .LBB11_5: # %cond.store3
+; AVX512FVL-NEXT: vpextrw $2, %xmm0, 4(%rdi)
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: je .LBB11_8
+; AVX512FVL-NEXT: .LBB11_7: # %cond.store5
+; AVX512FVL-NEXT: vpextrw $3, %xmm0, 6(%rdi)
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: je .LBB11_10
+; AVX512FVL-NEXT: .LBB11_9: # %cond.store7
+; AVX512FVL-NEXT: vpextrw $4, %xmm0, 8(%rdi)
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: je .LBB11_12
+; AVX512FVL-NEXT: .LBB11_11: # %cond.store9
+; AVX512FVL-NEXT: vpextrw $5, %xmm0, 10(%rdi)
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: je .LBB11_14
+; AVX512FVL-NEXT: .LBB11_13: # %cond.store11
+; AVX512FVL-NEXT: vpextrw $6, %xmm0, 12(%rdi)
+; AVX512FVL-NEXT: testb $-128, %al
+; AVX512FVL-NEXT: je .LBB11_16
+; AVX512FVL-NEXT: .LBB11_15: # %cond.store13
+; AVX512FVL-NEXT: vpextrw $7, %xmm0, 14(%rdi)
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v8i32_v8i16:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $ymm1 killed $ymm1 def $zmm1
@@ -3996,6 +4509,70 @@ define void @truncstore_v8i32_v8i8(<8 x i32> %x, ptr %p, <8 x i32> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v8i32_v8i8:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vptestmd %ymm1, %ymm1, %k0
+; AVX512FVL-NEXT: vpmovdb %ymm0, %xmm0
+; AVX512FVL-NEXT: kmovw %k0, %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB12_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB12_3
+; AVX512FVL-NEXT: .LBB12_4: # %else2
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: jne .LBB12_5
+; AVX512FVL-NEXT: .LBB12_6: # %else4
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: jne .LBB12_7
+; AVX512FVL-NEXT: .LBB12_8: # %else6
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: jne .LBB12_9
+; AVX512FVL-NEXT: .LBB12_10: # %else8
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: jne .LBB12_11
+; AVX512FVL-NEXT: .LBB12_12: # %else10
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: jne .LBB12_13
+; AVX512FVL-NEXT: .LBB12_14: # %else12
+; AVX512FVL-NEXT: testb $-128, %al
+; AVX512FVL-NEXT: jne .LBB12_15
+; AVX512FVL-NEXT: .LBB12_16: # %else14
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB12_1: # %cond.store
+; AVX512FVL-NEXT: vpextrb $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB12_4
+; AVX512FVL-NEXT: .LBB12_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrb $1, %xmm0, 1(%rdi)
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: je .LBB12_6
+; AVX512FVL-NEXT: .LBB12_5: # %cond.store3
+; AVX512FVL-NEXT: vpextrb $2, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: je .LBB12_8
+; AVX512FVL-NEXT: .LBB12_7: # %cond.store5
+; AVX512FVL-NEXT: vpextrb $3, %xmm0, 3(%rdi)
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: je .LBB12_10
+; AVX512FVL-NEXT: .LBB12_9: # %cond.store7
+; AVX512FVL-NEXT: vpextrb $4, %xmm0, 4(%rdi)
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: je .LBB12_12
+; AVX512FVL-NEXT: .LBB12_11: # %cond.store9
+; AVX512FVL-NEXT: vpextrb $5, %xmm0, 5(%rdi)
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: je .LBB12_14
+; AVX512FVL-NEXT: .LBB12_13: # %cond.store11
+; AVX512FVL-NEXT: vpextrb $6, %xmm0, 6(%rdi)
+; AVX512FVL-NEXT: testb $-128, %al
+; AVX512FVL-NEXT: je .LBB12_16
+; AVX512FVL-NEXT: .LBB12_15: # %cond.store13
+; AVX512FVL-NEXT: vpextrb $7, %xmm0, 7(%rdi)
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v8i32_v8i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $ymm1 killed $ymm1 def $zmm1
@@ -4172,6 +4749,40 @@ define void @truncstore_v4i32_v4i16(<4 x i32> %x, ptr %p, <4 x i32> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v4i32_v4i16:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vptestmd %xmm1, %xmm1, %k0
+; AVX512FVL-NEXT: vpmovdw %xmm0, %xmm0
+; AVX512FVL-NEXT: kmovw %k0, %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB13_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB13_3
+; AVX512FVL-NEXT: .LBB13_4: # %else2
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: jne .LBB13_5
+; AVX512FVL-NEXT: .LBB13_6: # %else4
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: jne .LBB13_7
+; AVX512FVL-NEXT: .LBB13_8: # %else6
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB13_1: # %cond.store
+; AVX512FVL-NEXT: vpextrw $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB13_4
+; AVX512FVL-NEXT: .LBB13_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrw $1, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: je .LBB13_6
+; AVX512FVL-NEXT: .LBB13_5: # %cond.store3
+; AVX512FVL-NEXT: vpextrw $2, %xmm0, 4(%rdi)
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: je .LBB13_8
+; AVX512FVL-NEXT: .LBB13_7: # %cond.store5
+; AVX512FVL-NEXT: vpextrw $3, %xmm0, 6(%rdi)
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v4i32_v4i16:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
@@ -4346,6 +4957,40 @@ define void @truncstore_v4i32_v4i8(<4 x i32> %x, ptr %p, <4 x i32> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v4i32_v4i8:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vptestmd %xmm1, %xmm1, %k0
+; AVX512FVL-NEXT: vpmovdb %xmm0, %xmm0
+; AVX512FVL-NEXT: kmovw %k0, %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB14_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB14_3
+; AVX512FVL-NEXT: .LBB14_4: # %else2
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: jne .LBB14_5
+; AVX512FVL-NEXT: .LBB14_6: # %else4
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: jne .LBB14_7
+; AVX512FVL-NEXT: .LBB14_8: # %else6
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB14_1: # %cond.store
+; AVX512FVL-NEXT: vpextrb $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB14_4
+; AVX512FVL-NEXT: .LBB14_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrb $1, %xmm0, 1(%rdi)
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: je .LBB14_6
+; AVX512FVL-NEXT: .LBB14_5: # %cond.store3
+; AVX512FVL-NEXT: vpextrb $2, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: je .LBB14_8
+; AVX512FVL-NEXT: .LBB14_7: # %cond.store5
+; AVX512FVL-NEXT: vpextrb $3, %xmm0, 3(%rdi)
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v4i32_v4i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
@@ -5550,6 +6195,245 @@ define void @truncstore_v32i16_v32i8(<32 x i16> %x, ptr %p, <32 x i8> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v32i16_v32i8:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vpxor %xmm2, %xmm2, %xmm2
+; AVX512FVL-NEXT: vpcmpeqb %ymm2, %ymm1, %ymm1
+; AVX512FVL-NEXT: vpmovzxwd {{.*#+}} zmm2 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
+; AVX512FVL-NEXT: vpmovdb %zmm2, %xmm2
+; AVX512FVL-NEXT: vextracti64x4 $1, %zmm0, %ymm0
+; AVX512FVL-NEXT: vpmovzxwd {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
+; AVX512FVL-NEXT: vpmovdb %zmm0, %xmm0
+; AVX512FVL-NEXT: vinserti128 $1, %xmm0, %ymm2, %ymm0
+; AVX512FVL-NEXT: vpmovmskb %ymm1, %eax
+; AVX512FVL-NEXT: notl %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB15_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB15_3
+; AVX512FVL-NEXT: .LBB15_4: # %else2
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: jne .LBB15_5
+; AVX512FVL-NEXT: .LBB15_6: # %else4
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: jne .LBB15_7
+; AVX512FVL-NEXT: .LBB15_8: # %else6
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: jne .LBB15_9
+; AVX512FVL-NEXT: .LBB15_10: # %else8
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: jne .LBB15_11
+; AVX512FVL-NEXT: .LBB15_12: # %else10
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: jne .LBB15_13
+; AVX512FVL-NEXT: .LBB15_14: # %else12
+; AVX512FVL-NEXT: testb %al, %al
+; AVX512FVL-NEXT: js .LBB15_15
+; AVX512FVL-NEXT: .LBB15_16: # %else14
+; AVX512FVL-NEXT: testl $256, %eax # imm = 0x100
+; AVX512FVL-NEXT: jne .LBB15_17
+; AVX512FVL-NEXT: .LBB15_18: # %else16
+; AVX512FVL-NEXT: testl $512, %eax # imm = 0x200
+; AVX512FVL-NEXT: jne .LBB15_19
+; AVX512FVL-NEXT: .LBB15_20: # %else18
+; AVX512FVL-NEXT: testl $1024, %eax # imm = 0x400
+; AVX512FVL-NEXT: jne .LBB15_21
+; AVX512FVL-NEXT: .LBB15_22: # %else20
+; AVX512FVL-NEXT: testl $2048, %eax # imm = 0x800
+; AVX512FVL-NEXT: jne .LBB15_23
+; AVX512FVL-NEXT: .LBB15_24: # %else22
+; AVX512FVL-NEXT: testl $4096, %eax # imm = 0x1000
+; AVX512FVL-NEXT: jne .LBB15_25
+; AVX512FVL-NEXT: .LBB15_26: # %else24
+; AVX512FVL-NEXT: testl $8192, %eax # imm = 0x2000
+; AVX512FVL-NEXT: jne .LBB15_27
+; AVX512FVL-NEXT: .LBB15_28: # %else26
+; AVX512FVL-NEXT: testl $16384, %eax # imm = 0x4000
+; AVX512FVL-NEXT: jne .LBB15_29
+; AVX512FVL-NEXT: .LBB15_30: # %else28
+; AVX512FVL-NEXT: testw %ax, %ax
+; AVX512FVL-NEXT: jns .LBB15_32
+; AVX512FVL-NEXT: .LBB15_31: # %cond.store29
+; AVX512FVL-NEXT: vpextrb $15, %xmm0, 15(%rdi)
+; AVX512FVL-NEXT: .LBB15_32: # %else30
+; AVX512FVL-NEXT: testl $65536, %eax # imm = 0x10000
+; AVX512FVL-NEXT: vextracti128 $1, %ymm0, %xmm0
+; AVX512FVL-NEXT: jne .LBB15_33
+; AVX512FVL-NEXT: # %bb.34: # %else32
+; AVX512FVL-NEXT: testl $131072, %eax # imm = 0x20000
+; AVX512FVL-NEXT: jne .LBB15_35
+; AVX512FVL-NEXT: .LBB15_36: # %else34
+; AVX512FVL-NEXT: testl $262144, %eax # imm = 0x40000
+; AVX512FVL-NEXT: jne .LBB15_37
+; AVX512FVL-NEXT: .LBB15_38: # %else36
+; AVX512FVL-NEXT: testl $524288, %eax # imm = 0x80000
+; AVX512FVL-NEXT: jne .LBB15_39
+; AVX512FVL-NEXT: .LBB15_40: # %else38
+; AVX512FVL-NEXT: testl $1048576, %eax # imm = 0x100000
+; AVX512FVL-NEXT: jne .LBB15_41
+; AVX512FVL-NEXT: .LBB15_42: # %else40
+; AVX512FVL-NEXT: testl $2097152, %eax # imm = 0x200000
+; AVX512FVL-NEXT: jne .LBB15_43
+; AVX512FVL-NEXT: .LBB15_44: # %else42
+; AVX512FVL-NEXT: testl $4194304, %eax # imm = 0x400000
+; AVX512FVL-NEXT: jne .LBB15_45
+; AVX512FVL-NEXT: .LBB15_46: # %else44
+; AVX512FVL-NEXT: testl $8388608, %eax # imm = 0x800000
+; AVX512FVL-NEXT: jne .LBB15_47
+; AVX512FVL-NEXT: .LBB15_48: # %else46
+; AVX512FVL-NEXT: testl $16777216, %eax # imm = 0x1000000
+; AVX512FVL-NEXT: jne .LBB15_49
+; AVX512FVL-NEXT: .LBB15_50: # %else48
+; AVX512FVL-NEXT: testl $33554432, %eax # imm = 0x2000000
+; AVX512FVL-NEXT: jne .LBB15_51
+; AVX512FVL-NEXT: .LBB15_52: # %else50
+; AVX512FVL-NEXT: testl $67108864, %eax # imm = 0x4000000
+; AVX512FVL-NEXT: jne .LBB15_53
+; AVX512FVL-NEXT: .LBB15_54: # %else52
+; AVX512FVL-NEXT: testl $134217728, %eax # imm = 0x8000000
+; AVX512FVL-NEXT: jne .LBB15_55
+; AVX512FVL-NEXT: .LBB15_56: # %else54
+; AVX512FVL-NEXT: testl $268435456, %eax # imm = 0x10000000
+; AVX512FVL-NEXT: jne .LBB15_57
+; AVX512FVL-NEXT: .LBB15_58: # %else56
+; AVX512FVL-NEXT: testl $536870912, %eax # imm = 0x20000000
+; AVX512FVL-NEXT: jne .LBB15_59
+; AVX512FVL-NEXT: .LBB15_60: # %else58
+; AVX512FVL-NEXT: testl $1073741824, %eax # imm = 0x40000000
+; AVX512FVL-NEXT: jne .LBB15_61
+; AVX512FVL-NEXT: .LBB15_62: # %else60
+; AVX512FVL-NEXT: testl $-2147483648, %eax # imm = 0x80000000
+; AVX512FVL-NEXT: jne .LBB15_63
+; AVX512FVL-NEXT: .LBB15_64: # %else62
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB15_1: # %cond.store
+; AVX512FVL-NEXT: vpextrb $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB15_4
+; AVX512FVL-NEXT: .LBB15_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrb $1, %xmm0, 1(%rdi)
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: je .LBB15_6
+; AVX512FVL-NEXT: .LBB15_5: # %cond.store3
+; AVX512FVL-NEXT: vpextrb $2, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: je .LBB15_8
+; AVX512FVL-NEXT: .LBB15_7: # %cond.store5
+; AVX512FVL-NEXT: vpextrb $3, %xmm0, 3(%rdi)
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: je .LBB15_10
+; AVX512FVL-NEXT: .LBB15_9: # %cond.store7
+; AVX512FVL-NEXT: vpextrb $4, %xmm0, 4(%rdi)
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: je .LBB15_12
+; AVX512FVL-NEXT: .LBB15_11: # %cond.store9
+; AVX512FVL-NEXT: vpextrb $5, %xmm0, 5(%rdi)
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: je .LBB15_14
+; AVX512FVL-NEXT: .LBB15_13: # %cond.store11
+; AVX512FVL-NEXT: vpextrb $6, %xmm0, 6(%rdi)
+; AVX512FVL-NEXT: testb %al, %al
+; AVX512FVL-NEXT: jns .LBB15_16
+; AVX512FVL-NEXT: .LBB15_15: # %cond.store13
+; AVX512FVL-NEXT: vpextrb $7, %xmm0, 7(%rdi)
+; AVX512FVL-NEXT: testl $256, %eax # imm = 0x100
+; AVX512FVL-NEXT: je .LBB15_18
+; AVX512FVL-NEXT: .LBB15_17: # %cond.store15
+; AVX512FVL-NEXT: vpextrb $8, %xmm0, 8(%rdi)
+; AVX512FVL-NEXT: testl $512, %eax # imm = 0x200
+; AVX512FVL-NEXT: je .LBB15_20
+; AVX512FVL-NEXT: .LBB15_19: # %cond.store17
+; AVX512FVL-NEXT: vpextrb $9, %xmm0, 9(%rdi)
+; AVX512FVL-NEXT: testl $1024, %eax # imm = 0x400
+; AVX512FVL-NEXT: je .LBB15_22
+; AVX512FVL-NEXT: .LBB15_21: # %cond.store19
+; AVX512FVL-NEXT: vpextrb $10, %xmm0, 10(%rdi)
+; AVX512FVL-NEXT: testl $2048, %eax # imm = 0x800
+; AVX512FVL-NEXT: je .LBB15_24
+; AVX512FVL-NEXT: .LBB15_23: # %cond.store21
+; AVX512FVL-NEXT: vpextrb $11, %xmm0, 11(%rdi)
+; AVX512FVL-NEXT: testl $4096, %eax # imm = 0x1000
+; AVX512FVL-NEXT: je .LBB15_26
+; AVX512FVL-NEXT: .LBB15_25: # %cond.store23
+; AVX512FVL-NEXT: vpextrb $12, %xmm0, 12(%rdi)
+; AVX512FVL-NEXT: testl $8192, %eax # imm = 0x2000
+; AVX512FVL-NEXT: je .LBB15_28
+; AVX512FVL-NEXT: .LBB15_27: # %cond.store25
+; AVX512FVL-NEXT: vpextrb $13, %xmm0, 13(%rdi)
+; AVX512FVL-NEXT: testl $16384, %eax # imm = 0x4000
+; AVX512FVL-NEXT: je .LBB15_30
+; AVX512FVL-NEXT: .LBB15_29: # %cond.store27
+; AVX512FVL-NEXT: vpextrb $14, %xmm0, 14(%rdi)
+; AVX512FVL-NEXT: testw %ax, %ax
+; AVX512FVL-NEXT: js .LBB15_31
+; AVX512FVL-NEXT: jmp .LBB15_32
+; AVX512FVL-NEXT: .LBB15_33: # %cond.store31
+; AVX512FVL-NEXT: vpextrb $0, %xmm0, 16(%rdi)
+; AVX512FVL-NEXT: testl $131072, %eax # imm = 0x20000
+; AVX512FVL-NEXT: je .LBB15_36
+; AVX512FVL-NEXT: .LBB15_35: # %cond.store33
+; AVX512FVL-NEXT: vpextrb $1, %xmm0, 17(%rdi)
+; AVX512FVL-NEXT: testl $262144, %eax # imm = 0x40000
+; AVX512FVL-NEXT: je .LBB15_38
+; AVX512FVL-NEXT: .LBB15_37: # %cond.store35
+; AVX512FVL-NEXT: vpextrb $2, %xmm0, 18(%rdi)
+; AVX512FVL-NEXT: testl $524288, %eax # imm = 0x80000
+; AVX512FVL-NEXT: je .LBB15_40
+; AVX512FVL-NEXT: .LBB15_39: # %cond.store37
+; AVX512FVL-NEXT: vpextrb $3, %xmm0, 19(%rdi)
+; AVX512FVL-NEXT: testl $1048576, %eax # imm = 0x100000
+; AVX512FVL-NEXT: je .LBB15_42
+; AVX512FVL-NEXT: .LBB15_41: # %cond.store39
+; AVX512FVL-NEXT: vpextrb $4, %xmm0, 20(%rdi)
+; AVX512FVL-NEXT: testl $2097152, %eax # imm = 0x200000
+; AVX512FVL-NEXT: je .LBB15_44
+; AVX512FVL-NEXT: .LBB15_43: # %cond.store41
+; AVX512FVL-NEXT: vpextrb $5, %xmm0, 21(%rdi)
+; AVX512FVL-NEXT: testl $4194304, %eax # imm = 0x400000
+; AVX512FVL-NEXT: je .LBB15_46
+; AVX512FVL-NEXT: .LBB15_45: # %cond.store43
+; AVX512FVL-NEXT: vpextrb $6, %xmm0, 22(%rdi)
+; AVX512FVL-NEXT: testl $8388608, %eax # imm = 0x800000
+; AVX512FVL-NEXT: je .LBB15_48
+; AVX512FVL-NEXT: .LBB15_47: # %cond.store45
+; AVX512FVL-NEXT: vpextrb $7, %xmm0, 23(%rdi)
+; AVX512FVL-NEXT: testl $16777216, %eax # imm = 0x1000000
+; AVX512FVL-NEXT: je .LBB15_50
+; AVX512FVL-NEXT: .LBB15_49: # %cond.store47
+; AVX512FVL-NEXT: vpextrb $8, %xmm0, 24(%rdi)
+; AVX512FVL-NEXT: testl $33554432, %eax # imm = 0x2000000
+; AVX512FVL-NEXT: je .LBB15_52
+; AVX512FVL-NEXT: .LBB15_51: # %cond.store49
+; AVX512FVL-NEXT: vpextrb $9, %xmm0, 25(%rdi)
+; AVX512FVL-NEXT: testl $67108864, %eax # imm = 0x4000000
+; AVX512FVL-NEXT: je .LBB15_54
+; AVX512FVL-NEXT: .LBB15_53: # %cond.store51
+; AVX512FVL-NEXT: vpextrb $10, %xmm0, 26(%rdi)
+; AVX512FVL-NEXT: testl $134217728, %eax # imm = 0x8000000
+; AVX512FVL-NEXT: je .LBB15_56
+; AVX512FVL-NEXT: .LBB15_55: # %cond.store53
+; AVX512FVL-NEXT: vpextrb $11, %xmm0, 27(%rdi)
+; AVX512FVL-NEXT: testl $268435456, %eax # imm = 0x10000000
+; AVX512FVL-NEXT: je .LBB15_58
+; AVX512FVL-NEXT: .LBB15_57: # %cond.store55
+; AVX512FVL-NEXT: vpextrb $12, %xmm0, 28(%rdi)
+; AVX512FVL-NEXT: testl $536870912, %eax # imm = 0x20000000
+; AVX512FVL-NEXT: je .LBB15_60
+; AVX512FVL-NEXT: .LBB15_59: # %cond.store57
+; AVX512FVL-NEXT: vpextrb $13, %xmm0, 29(%rdi)
+; AVX512FVL-NEXT: testl $1073741824, %eax # imm = 0x40000000
+; AVX512FVL-NEXT: je .LBB15_62
+; AVX512FVL-NEXT: .LBB15_61: # %cond.store59
+; AVX512FVL-NEXT: vpextrb $14, %xmm0, 30(%rdi)
+; AVX512FVL-NEXT: testl $-2147483648, %eax # imm = 0x80000000
+; AVX512FVL-NEXT: je .LBB15_64
+; AVX512FVL-NEXT: .LBB15_63: # %cond.store61
+; AVX512FVL-NEXT: vpextrb $15, %xmm0, 31(%rdi)
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v32i16_v32i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $ymm1 killed $ymm1 def $zmm1
@@ -6177,6 +7061,129 @@ define void @truncstore_v16i16_v16i8(<16 x i16> %x, ptr %p, <16 x i8> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v16i16_v16i8:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vpxor %xmm2, %xmm2, %xmm2
+; AVX512FVL-NEXT: vpcmpeqb %xmm2, %xmm1, %xmm1
+; AVX512FVL-NEXT: vpmovzxwd {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
+; AVX512FVL-NEXT: vpmovdb %zmm0, %xmm0
+; AVX512FVL-NEXT: vpmovmskb %xmm1, %eax
+; AVX512FVL-NEXT: xorl $65535, %eax # imm = 0xFFFF
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB16_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB16_3
+; AVX512FVL-NEXT: .LBB16_4: # %else2
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: jne .LBB16_5
+; AVX512FVL-NEXT: .LBB16_6: # %else4
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: jne .LBB16_7
+; AVX512FVL-NEXT: .LBB16_8: # %else6
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: jne .LBB16_9
+; AVX512FVL-NEXT: .LBB16_10: # %else8
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: jne .LBB16_11
+; AVX512FVL-NEXT: .LBB16_12: # %else10
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: jne .LBB16_13
+; AVX512FVL-NEXT: .LBB16_14: # %else12
+; AVX512FVL-NEXT: testb %al, %al
+; AVX512FVL-NEXT: js .LBB16_15
+; AVX512FVL-NEXT: .LBB16_16: # %else14
+; AVX512FVL-NEXT: testl $256, %eax # imm = 0x100
+; AVX512FVL-NEXT: jne .LBB16_17
+; AVX512FVL-NEXT: .LBB16_18: # %else16
+; AVX512FVL-NEXT: testl $512, %eax # imm = 0x200
+; AVX512FVL-NEXT: jne .LBB16_19
+; AVX512FVL-NEXT: .LBB16_20: # %else18
+; AVX512FVL-NEXT: testl $1024, %eax # imm = 0x400
+; AVX512FVL-NEXT: jne .LBB16_21
+; AVX512FVL-NEXT: .LBB16_22: # %else20
+; AVX512FVL-NEXT: testl $2048, %eax # imm = 0x800
+; AVX512FVL-NEXT: jne .LBB16_23
+; AVX512FVL-NEXT: .LBB16_24: # %else22
+; AVX512FVL-NEXT: testl $4096, %eax # imm = 0x1000
+; AVX512FVL-NEXT: jne .LBB16_25
+; AVX512FVL-NEXT: .LBB16_26: # %else24
+; AVX512FVL-NEXT: testl $8192, %eax # imm = 0x2000
+; AVX512FVL-NEXT: jne .LBB16_27
+; AVX512FVL-NEXT: .LBB16_28: # %else26
+; AVX512FVL-NEXT: testl $16384, %eax # imm = 0x4000
+; AVX512FVL-NEXT: jne .LBB16_29
+; AVX512FVL-NEXT: .LBB16_30: # %else28
+; AVX512FVL-NEXT: testl $32768, %eax # imm = 0x8000
+; AVX512FVL-NEXT: jne .LBB16_31
+; AVX512FVL-NEXT: .LBB16_32: # %else30
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB16_1: # %cond.store
+; AVX512FVL-NEXT: vpextrb $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB16_4
+; AVX512FVL-NEXT: .LBB16_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrb $1, %xmm0, 1(%rdi)
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: je .LBB16_6
+; AVX512FVL-NEXT: .LBB16_5: # %cond.store3
+; AVX512FVL-NEXT: vpextrb $2, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: je .LBB16_8
+; AVX512FVL-NEXT: .LBB16_7: # %cond.store5
+; AVX512FVL-NEXT: vpextrb $3, %xmm0, 3(%rdi)
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: je .LBB16_10
+; AVX512FVL-NEXT: .LBB16_9: # %cond.store7
+; AVX512FVL-NEXT: vpextrb $4, %xmm0, 4(%rdi)
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: je .LBB16_12
+; AVX512FVL-NEXT: .LBB16_11: # %cond.store9
+; AVX512FVL-NEXT: vpextrb $5, %xmm0, 5(%rdi)
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: je .LBB16_14
+; AVX512FVL-NEXT: .LBB16_13: # %cond.store11
+; AVX512FVL-NEXT: vpextrb $6, %xmm0, 6(%rdi)
+; AVX512FVL-NEXT: testb %al, %al
+; AVX512FVL-NEXT: jns .LBB16_16
+; AVX512FVL-NEXT: .LBB16_15: # %cond.store13
+; AVX512FVL-NEXT: vpextrb $7, %xmm0, 7(%rdi)
+; AVX512FVL-NEXT: testl $256, %eax # imm = 0x100
+; AVX512FVL-NEXT: je .LBB16_18
+; AVX512FVL-NEXT: .LBB16_17: # %cond.store15
+; AVX512FVL-NEXT: vpextrb $8, %xmm0, 8(%rdi)
+; AVX512FVL-NEXT: testl $512, %eax # imm = 0x200
+; AVX512FVL-NEXT: je .LBB16_20
+; AVX512FVL-NEXT: .LBB16_19: # %cond.store17
+; AVX512FVL-NEXT: vpextrb $9, %xmm0, 9(%rdi)
+; AVX512FVL-NEXT: testl $1024, %eax # imm = 0x400
+; AVX512FVL-NEXT: je .LBB16_22
+; AVX512FVL-NEXT: .LBB16_21: # %cond.store19
+; AVX512FVL-NEXT: vpextrb $10, %xmm0, 10(%rdi)
+; AVX512FVL-NEXT: testl $2048, %eax # imm = 0x800
+; AVX512FVL-NEXT: je .LBB16_24
+; AVX512FVL-NEXT: .LBB16_23: # %cond.store21
+; AVX512FVL-NEXT: vpextrb $11, %xmm0, 11(%rdi)
+; AVX512FVL-NEXT: testl $4096, %eax # imm = 0x1000
+; AVX512FVL-NEXT: je .LBB16_26
+; AVX512FVL-NEXT: .LBB16_25: # %cond.store23
+; AVX512FVL-NEXT: vpextrb $12, %xmm0, 12(%rdi)
+; AVX512FVL-NEXT: testl $8192, %eax # imm = 0x2000
+; AVX512FVL-NEXT: je .LBB16_28
+; AVX512FVL-NEXT: .LBB16_27: # %cond.store25
+; AVX512FVL-NEXT: vpextrb $13, %xmm0, 13(%rdi)
+; AVX512FVL-NEXT: testl $16384, %eax # imm = 0x4000
+; AVX512FVL-NEXT: je .LBB16_30
+; AVX512FVL-NEXT: .LBB16_29: # %cond.store27
+; AVX512FVL-NEXT: vpextrb $14, %xmm0, 14(%rdi)
+; AVX512FVL-NEXT: testl $32768, %eax # imm = 0x8000
+; AVX512FVL-NEXT: je .LBB16_32
+; AVX512FVL-NEXT: .LBB16_31: # %cond.store29
+; AVX512FVL-NEXT: vpextrb $15, %xmm0, 15(%rdi)
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v16i16_v16i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
@@ -6466,6 +7473,74 @@ define void @truncstore_v8i16_v8i8(<8 x i16> %x, ptr %p, <8 x i16> %mask) {
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
+; AVX512FVL-LABEL: truncstore_v8i16_v8i8:
+; AVX512FVL: # %bb.0:
+; AVX512FVL-NEXT: vpxor %xmm2, %xmm2, %xmm2
+; AVX512FVL-NEXT: vpcmpeqw %xmm2, %xmm1, %xmm1
+; AVX512FVL-NEXT: vpternlogq {{.*#+}} xmm1 = ~xmm1
+; AVX512FVL-NEXT: vpmovsxwd %xmm1, %ymm1
+; AVX512FVL-NEXT: vptestmd %ymm1, %ymm1, %k0
+; AVX512FVL-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX512FVL-NEXT: kmovw %k0, %eax
+; AVX512FVL-NEXT: testb $1, %al
+; AVX512FVL-NEXT: jne .LBB17_1
+; AVX512FVL-NEXT: # %bb.2: # %else
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: jne .LBB17_3
+; AVX512FVL-NEXT: .LBB17_4: # %else2
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: jne .LBB17_5
+; AVX512FVL-NEXT: .LBB17_6: # %else4
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: jne .LBB17_7
+; AVX512FVL-NEXT: .LBB17_8: # %else6
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: jne .LBB17_9
+; AVX512FVL-NEXT: .LBB17_10: # %else8
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: jne .LBB17_11
+; AVX512FVL-NEXT: .LBB17_12: # %else10
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: jne .LBB17_13
+; AVX512FVL-NEXT: .LBB17_14: # %else12
+; AVX512FVL-NEXT: testb $-128, %al
+; AVX512FVL-NEXT: jne .LBB17_15
+; AVX512FVL-NEXT: .LBB17_16: # %else14
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+; AVX512FVL-NEXT: .LBB17_1: # %cond.store
+; AVX512FVL-NEXT: vpextrb $0, %xmm0, (%rdi)
+; AVX512FVL-NEXT: testb $2, %al
+; AVX512FVL-NEXT: je .LBB17_4
+; AVX512FVL-NEXT: .LBB17_3: # %cond.store1
+; AVX512FVL-NEXT: vpextrb $1, %xmm0, 1(%rdi)
+; AVX512FVL-NEXT: testb $4, %al
+; AVX512FVL-NEXT: je .LBB17_6
+; AVX512FVL-NEXT: .LBB17_5: # %cond.store3
+; AVX512FVL-NEXT: vpextrb $2, %xmm0, 2(%rdi)
+; AVX512FVL-NEXT: testb $8, %al
+; AVX512FVL-NEXT: je .LBB17_8
+; AVX512FVL-NEXT: .LBB17_7: # %cond.store5
+; AVX512FVL-NEXT: vpextrb $3, %xmm0, 3(%rdi)
+; AVX512FVL-NEXT: testb $16, %al
+; AVX512FVL-NEXT: je .LBB17_10
+; AVX512FVL-NEXT: .LBB17_9: # %cond.store7
+; AVX512FVL-NEXT: vpextrb $4, %xmm0, 4(%rdi)
+; AVX512FVL-NEXT: testb $32, %al
+; AVX512FVL-NEXT: je .LBB17_12
+; AVX512FVL-NEXT: .LBB17_11: # %cond.store9
+; AVX512FVL-NEXT: vpextrb $5, %xmm0, 5(%rdi)
+; AVX512FVL-NEXT: testb $64, %al
+; AVX512FVL-NEXT: je .LBB17_14
+; AVX512FVL-NEXT: .LBB17_13: # %cond.store11
+; AVX512FVL-NEXT: vpextrb $6, %xmm0, 6(%rdi)
+; AVX512FVL-NEXT: testb $-128, %al
+; AVX512FVL-NEXT: je .LBB17_16
+; AVX512FVL-NEXT: .LBB17_15: # %cond.store13
+; AVX512FVL-NEXT: vpextrb $7, %xmm0, 7(%rdi)
+; AVX512FVL-NEXT: vzeroupper
+; AVX512FVL-NEXT: retq
+;
; AVX512BW-LABEL: truncstore_v8i16_v8i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
More information about the llvm-commits
mailing list