Skip to content

Commit 7bced74

Browse files
RKSimonRKSimon
authored
[X86] combine-icmp.ll - fix copy+paste typo in concat_icmp_v64i8_v16i8 test (#170281)
I changed the condcode for variety but failed to update the constant to prevent constant folding
1 parent 9ba5fa2 commit 7bced74

File tree

1 file changed

+84
-12
lines changed

1 file changed

+84
-12
lines changed

‎llvm/test/CodeGen/X86/combine-icmp.ll‎

Lines changed: 84 additions & 12 deletions
Original file line numberDiff line numberDiff line change
@@ -1,9 +1,9 @@
11
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
22
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 | FileCheck %s --check-prefixes=SSE,SSE2
33
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v2 | FileCheck %s --check-prefixes=SSE,SSE42
4-
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=sandybridge | FileCheck %s --check-prefixes=AVX,AVX1OR2,AVX1
5-
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v3 | FileCheck %s --check-prefixes=AVX,AVX1OR2,AVX2
6-
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v4 | FileCheck %s --check-prefixes=AVX,AVX512
4+
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=sandybridge | FileCheck %s --check-prefixes=AVX1OR2,AVX1
5+
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v3 | FileCheck %s --check-prefixes=AVX1OR2,AVX2
6+
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v4 | FileCheck %s --check-prefixes=AVX512
77

88
define i4 @concat_icmp_v4i64_v2i64(<2 x i64> %a0, <2 x i64> %a1) {
99
; SSE2-LABEL: concat_icmp_v4i64_v2i64:
@@ -491,17 +491,89 @@ define i32 @concat_icmp_v32i16_v8i16(<8 x i16> %a0, <8 x i16> %a1, <8 x i16> %a2
491491
define i64 @concat_icmp_v64i8_v16i8(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> %a2, <16 x i8> %a3) {
492492
; SSE-LABEL: concat_icmp_v64i8_v16i8:
493493
; SSE: # %bb.0:
494-
; SSE-NEXT: movq $-1, %rax
494+
; SSE-NEXT: movdqa {{.*#+}} xmm4 = [16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
495+
; SSE-NEXT: movdqa %xmm0, %xmm5
496+
; SSE-NEXT: pmaxub %xmm4, %xmm5
497+
; SSE-NEXT: pcmpeqb %xmm0, %xmm5
498+
; SSE-NEXT: pmovmskb %xmm5, %eax
499+
; SSE-NEXT: movdqa %xmm1, %xmm0
500+
; SSE-NEXT: pmaxub %xmm4, %xmm0
501+
; SSE-NEXT: pcmpeqb %xmm1, %xmm0
502+
; SSE-NEXT: pmovmskb %xmm0, %ecx
503+
; SSE-NEXT: shll $16, %ecx
504+
; SSE-NEXT: orl %eax, %ecx
505+
; SSE-NEXT: movdqa %xmm2, %xmm0
506+
; SSE-NEXT: pmaxub %xmm4, %xmm0
507+
; SSE-NEXT: pcmpeqb %xmm2, %xmm0
508+
; SSE-NEXT: pmovmskb %xmm0, %edx
509+
; SSE-NEXT: pmaxub %xmm3, %xmm4
510+
; SSE-NEXT: pcmpeqb %xmm3, %xmm4
511+
; SSE-NEXT: pmovmskb %xmm4, %eax
512+
; SSE-NEXT: shll $16, %eax
513+
; SSE-NEXT: orl %edx, %eax
514+
; SSE-NEXT: shlq $32, %rax
515+
; SSE-NEXT: orq %rcx, %rax
495516
; SSE-NEXT: retq
496517
;
497-
; AVX-LABEL: concat_icmp_v64i8_v16i8:
498-
; AVX: # %bb.0:
499-
; AVX-NEXT: movq $-1, %rax
500-
; AVX-NEXT: retq
501-
%v0 = icmp uge <16 x i8> %a0, zeroinitializer
502-
%v1 = icmp uge <16 x i8> %a1, zeroinitializer
503-
%v2 = icmp uge <16 x i8> %a2, zeroinitializer
504-
%v3 = icmp uge <16 x i8> %a3, zeroinitializer
518+
; AVX1-LABEL: concat_icmp_v64i8_v16i8:
519+
; AVX1: # %bb.0:
520+
; AVX1-NEXT: vbroadcastss {{.*#+}} xmm4 = [16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
521+
; AVX1-NEXT: vpmaxub %xmm4, %xmm0, %xmm5
522+
; AVX1-NEXT: vpcmpeqb %xmm5, %xmm0, %xmm0
523+
; AVX1-NEXT: vpmaxub %xmm4, %xmm1, %xmm5
524+
; AVX1-NEXT: vpcmpeqb %xmm5, %xmm1, %xmm1
525+
; AVX1-NEXT: vpmaxub %xmm4, %xmm2, %xmm5
526+
; AVX1-NEXT: vpcmpeqb %xmm5, %xmm2, %xmm2
527+
; AVX1-NEXT: vpmaxub %xmm4, %xmm3, %xmm4
528+
; AVX1-NEXT: vpcmpeqb %xmm4, %xmm3, %xmm3
529+
; AVX1-NEXT: vpmovmskb %xmm0, %eax
530+
; AVX1-NEXT: vpmovmskb %xmm1, %ecx
531+
; AVX1-NEXT: shll $16, %ecx
532+
; AVX1-NEXT: orl %eax, %ecx
533+
; AVX1-NEXT: vpmovmskb %xmm2, %edx
534+
; AVX1-NEXT: vpmovmskb %xmm3, %eax
535+
; AVX1-NEXT: shll $16, %eax
536+
; AVX1-NEXT: orl %edx, %eax
537+
; AVX1-NEXT: shlq $32, %rax
538+
; AVX1-NEXT: orq %rcx, %rax
539+
; AVX1-NEXT: retq
540+
;
541+
; AVX2-LABEL: concat_icmp_v64i8_v16i8:
542+
; AVX2: # %bb.0:
543+
; AVX2-NEXT: vpbroadcastd {{.*#+}} xmm4 = [16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
544+
; AVX2-NEXT: vpmaxub %xmm4, %xmm0, %xmm5
545+
; AVX2-NEXT: vpcmpeqb %xmm5, %xmm0, %xmm0
546+
; AVX2-NEXT: vpmaxub %xmm4, %xmm1, %xmm5
547+
; AVX2-NEXT: vpcmpeqb %xmm5, %xmm1, %xmm1
548+
; AVX2-NEXT: vpmaxub %xmm4, %xmm2, %xmm5
549+
; AVX2-NEXT: vpcmpeqb %xmm5, %xmm2, %xmm2
550+
; AVX2-NEXT: vpmaxub %xmm4, %xmm3, %xmm4
551+
; AVX2-NEXT: vpcmpeqb %xmm4, %xmm3, %xmm3
552+
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
553+
; AVX2-NEXT: vpmovmskb %ymm0, %ecx
554+
; AVX2-NEXT: vinserti128 $1, %xmm3, %ymm2, %ymm0
555+
; AVX2-NEXT: vpmovmskb %ymm0, %eax
556+
; AVX2-NEXT: shlq $32, %rax
557+
; AVX2-NEXT: orq %rcx, %rax
558+
; AVX2-NEXT: vzeroupper
559+
; AVX2-NEXT: retq
560+
;
561+
; AVX512-LABEL: concat_icmp_v64i8_v16i8:
562+
; AVX512: # %bb.0:
563+
; AVX512-NEXT: vpbroadcastd {{.*#+}} xmm4 = [15,15,15,15,15,15,15,15,15,15,15,15,15,15,15,15]
564+
; AVX512-NEXT: vpcmpnleub %xmm4, %xmm0, %k0
565+
; AVX512-NEXT: vpcmpnleub %xmm4, %xmm1, %k1
566+
; AVX512-NEXT: vpcmpnleub %xmm4, %xmm2, %k2
567+
; AVX512-NEXT: vpcmpnleub %xmm4, %xmm3, %k3
568+
; AVX512-NEXT: kunpckwd %k0, %k1, %k0
569+
; AVX512-NEXT: kunpckwd %k2, %k3, %k1
570+
; AVX512-NEXT: kunpckdq %k0, %k1, %k0
571+
; AVX512-NEXT: kmovq %k0, %rax
572+
; AVX512-NEXT: retq
573+
%v0 = icmp ugt <16 x i8> %a0, splat (i8 15)
574+
%v1 = icmp ugt <16 x i8> %a1, splat (i8 15)
575+
%v2 = icmp ugt <16 x i8> %a2, splat (i8 15)
576+
%v3 = icmp ugt <16 x i8> %a3, splat (i8 15)
505577
%v01 = shufflevector <16 x i1> %v0, <16 x i1> %v1, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
506578
%v23 = shufflevector <16 x i1> %v2, <16 x i1> %v3, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
507579
%v = shufflevector <32 x i1> %v01, <32 x i1> %v23, <64 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>

0 commit comments

Comments
 (0)