Skip to content

[X86] Add tests showing failure to concat fcmp instructions together #170313

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Dec 2, 2025
Merged

[X86] Add tests showing failure to concat fcmp instructions together #170313

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
330 changes: 330 additions & 0 deletions llvm/test/CodeGen/X86/combine-fcmp.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,330 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 | FileCheck %s --check-prefixes=SSE
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v2 | FileCheck %s --check-prefixes=SSE
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=sandybridge | FileCheck %s --check-prefixes=AVX1OR2,AVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v3 | FileCheck %s --check-prefixes=AVX1OR2,AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64-v4 | FileCheck %s --check-prefixes=AVX512

define i4 @concat_fcmp_v4f64_v2f64(<2 x double> %a0, <2 x double> %a1) {
; SSE-LABEL: concat_fcmp_v4f64_v2f64:
; SSE: # %bb.0:
; SSE-NEXT: xorpd %xmm2, %xmm2
; SSE-NEXT: xorpd %xmm3, %xmm3
; SSE-NEXT: cmpltpd %xmm0, %xmm3
; SSE-NEXT: cmpltpd %xmm1, %xmm2
; SSE-NEXT: shufps {{.*#+}} xmm3 = xmm3[0,2],xmm2[0,2]
; SSE-NEXT: movmskps %xmm3, %eax
; SSE-NEXT: # kill: def $al killed $al killed $eax
; SSE-NEXT: retq
;
; AVX1OR2-LABEL: concat_fcmp_v4f64_v2f64:
; AVX1OR2: # %bb.0:
; AVX1OR2-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; AVX1OR2-NEXT: vcmpltpd %xmm0, %xmm2, %xmm0
; AVX1OR2-NEXT: vcmpltpd %xmm1, %xmm2, %xmm1
; AVX1OR2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
; AVX1OR2-NEXT: vmovmskps %xmm0, %eax
; AVX1OR2-NEXT: # kill: def $al killed $al killed $eax
; AVX1OR2-NEXT: retq
;
; AVX512-LABEL: concat_fcmp_v4f64_v2f64:
; AVX512: # %bb.0:
; AVX512-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; AVX512-NEXT: vcmpltpd %xmm0, %xmm2, %k0
; AVX512-NEXT: vcmpltpd %xmm1, %xmm2, %k1
; AVX512-NEXT: kshiftlb $2, %k1, %k1
; AVX512-NEXT: korw %k1, %k0, %k0
; AVX512-NEXT: kmovd %k0, %eax
; AVX512-NEXT: # kill: def $al killed $al killed $eax
; AVX512-NEXT: retq
%v0 = fcmp ogt <2 x double> %a0, zeroinitializer
%v1 = fcmp ogt <2 x double> %a1, zeroinitializer
%v = shufflevector <2 x i1> %v0, <2 x i1> %v1, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
%r = bitcast <4 x i1> %v to i4
ret i4 %r
}

define i8 @concat_fcmp_v8f32_v4f32(<4 x float> %a0, <4 x float> %a1) {
; SSE-LABEL: concat_fcmp_v8f32_v4f32:
; SSE: # %bb.0:
; SSE-NEXT: xorps %xmm2, %xmm2
; SSE-NEXT: cmpeqps %xmm2, %xmm0
; SSE-NEXT: cmpeqps %xmm2, %xmm1
; SSE-NEXT: packssdw %xmm1, %xmm0
; SSE-NEXT: packsswb %xmm0, %xmm0
; SSE-NEXT: pmovmskb %xmm0, %eax
; SSE-NEXT: # kill: def $al killed $al killed $eax
; SSE-NEXT: retq
;
; AVX1OR2-LABEL: concat_fcmp_v8f32_v4f32:
; AVX1OR2: # %bb.0:
; AVX1OR2-NEXT: vxorps %xmm2, %xmm2, %xmm2
; AVX1OR2-NEXT: vcmpeqps %xmm2, %xmm0, %xmm0
; AVX1OR2-NEXT: vcmpeqps %xmm2, %xmm1, %xmm1
; AVX1OR2-NEXT: vpackssdw %xmm1, %xmm0, %xmm0
; AVX1OR2-NEXT: vpacksswb %xmm0, %xmm0, %xmm0
; AVX1OR2-NEXT: vpmovmskb %xmm0, %eax
; AVX1OR2-NEXT: # kill: def $al killed $al killed $eax
; AVX1OR2-NEXT: retq
;
; AVX512-LABEL: concat_fcmp_v8f32_v4f32:
; AVX512: # %bb.0:
; AVX512-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX512-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX512-NEXT: vxorps %xmm1, %xmm1, %xmm1
; AVX512-NEXT: vcmpeqps %ymm1, %ymm0, %k0
; AVX512-NEXT: kmovd %k0, %eax
; AVX512-NEXT: # kill: def $al killed $al killed $eax
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%v0 = fcmp oeq <4 x float> %a0, zeroinitializer
%v1 = fcmp oeq <4 x float> %a1, zeroinitializer
%v = shufflevector <4 x i1> %v0, <4 x i1> %v1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
%r = bitcast <8 x i1> %v to i8
ret i8 %r
}

define i8 @concat_fcmp_v8f64_v2f64(<2 x double> %a0, <2 x double> %a1, <2 x double> %a2, <2 x double> %a3) {
; SSE-LABEL: concat_fcmp_v8f64_v2f64:
; SSE: # %bb.0:
; SSE-NEXT: xorpd %xmm4, %xmm4
; SSE-NEXT: cmpltpd %xmm4, %xmm0
; SSE-NEXT: cmpltpd %xmm4, %xmm1
; SSE-NEXT: packssdw %xmm1, %xmm0
; SSE-NEXT: cmpltpd %xmm4, %xmm2
; SSE-NEXT: cmpltpd %xmm4, %xmm3
; SSE-NEXT: packssdw %xmm3, %xmm2
; SSE-NEXT: packssdw %xmm0, %xmm0
; SSE-NEXT: packssdw %xmm2, %xmm2
; SSE-NEXT: packsswb %xmm2, %xmm0
; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,3,2,3]
; SSE-NEXT: pmovmskb %xmm0, %eax
; SSE-NEXT: # kill: def $al killed $al killed $eax
; SSE-NEXT: retq
;
; AVX1OR2-LABEL: concat_fcmp_v8f64_v2f64:
; AVX1OR2: # %bb.0:
; AVX1OR2-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; AVX1OR2-NEXT: vcmpltpd %xmm4, %xmm0, %xmm0
; AVX1OR2-NEXT: vcmpltpd %xmm4, %xmm1, %xmm1
; AVX1OR2-NEXT: vpackssdw %xmm1, %xmm0, %xmm0
; AVX1OR2-NEXT: vcmpltpd %xmm4, %xmm2, %xmm1
; AVX1OR2-NEXT: vcmpltpd %xmm4, %xmm3, %xmm2
; AVX1OR2-NEXT: vpackssdw %xmm2, %xmm1, %xmm1
; AVX1OR2-NEXT: vpackssdw %xmm1, %xmm1, %xmm1
; AVX1OR2-NEXT: vpackssdw %xmm0, %xmm0, %xmm0
; AVX1OR2-NEXT: vpacksswb %xmm1, %xmm0, %xmm0
; AVX1OR2-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,3,0,3]
; AVX1OR2-NEXT: vpmovmskb %xmm0, %eax
; AVX1OR2-NEXT: # kill: def $al killed $al killed $eax
; AVX1OR2-NEXT: retq
;
; AVX512-LABEL: concat_fcmp_v8f64_v2f64:
; AVX512: # %bb.0:
; AVX512-NEXT: # kill: def $xmm2 killed $xmm2 def $ymm2
; AVX512-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX512-NEXT: vinsertf128 $1, %xmm3, %ymm2, %ymm2
; AVX512-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX512-NEXT: vinsertf64x4 $1, %ymm2, %zmm0, %zmm0
; AVX512-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; AVX512-NEXT: vcmpltpd %zmm1, %zmm0, %k0
; AVX512-NEXT: kmovd %k0, %eax
; AVX512-NEXT: # kill: def $al killed $al killed $eax
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%v0 = fcmp olt <2 x double> %a0, zeroinitializer
%v1 = fcmp olt <2 x double> %a1, zeroinitializer
%v2 = fcmp olt <2 x double> %a2, zeroinitializer
%v3 = fcmp olt <2 x double> %a3, zeroinitializer
%v01 = shufflevector <2 x i1> %v0, <2 x i1> %v1, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
%v23 = shufflevector <2 x i1> %v2, <2 x i1> %v3, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
%v = shufflevector <4 x i1> %v01, <4 x i1> %v23, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
%r = bitcast <8 x i1> %v to i8
ret i8 %r
}

define i16 @concat_fcmp_v16f32_v4f32(<4 x float> %a0, <4 x float> %a1, <4 x float> %a2, <4 x float> %a3) {
; SSE-LABEL: concat_fcmp_v16f32_v4f32:
; SSE: # %bb.0:
; SSE-NEXT: xorps %xmm4, %xmm4
; SSE-NEXT: xorps %xmm5, %xmm5
; SSE-NEXT: cmpleps %xmm0, %xmm5
; SSE-NEXT: xorps %xmm0, %xmm0
; SSE-NEXT: cmpleps %xmm1, %xmm0
; SSE-NEXT: packssdw %xmm0, %xmm5
; SSE-NEXT: xorps %xmm0, %xmm0
; SSE-NEXT: cmpleps %xmm2, %xmm0
; SSE-NEXT: cmpleps %xmm3, %xmm4
; SSE-NEXT: packssdw %xmm4, %xmm0
; SSE-NEXT: packsswb %xmm0, %xmm5
; SSE-NEXT: pmovmskb %xmm5, %eax
; SSE-NEXT: # kill: def $ax killed $ax killed $eax
; SSE-NEXT: retq
;
; AVX1OR2-LABEL: concat_fcmp_v16f32_v4f32:
; AVX1OR2: # %bb.0:
; AVX1OR2-NEXT: vxorps %xmm4, %xmm4, %xmm4
; AVX1OR2-NEXT: vcmpleps %xmm0, %xmm4, %xmm0
; AVX1OR2-NEXT: vcmpleps %xmm1, %xmm4, %xmm1
; AVX1OR2-NEXT: vpackssdw %xmm1, %xmm0, %xmm0
; AVX1OR2-NEXT: vcmpleps %xmm2, %xmm4, %xmm1
; AVX1OR2-NEXT: vcmpleps %xmm3, %xmm4, %xmm2
; AVX1OR2-NEXT: vpackssdw %xmm2, %xmm1, %xmm1
; AVX1OR2-NEXT: vpacksswb %xmm1, %xmm0, %xmm0
; AVX1OR2-NEXT: vpmovmskb %xmm0, %eax
; AVX1OR2-NEXT: # kill: def $ax killed $ax killed $eax
; AVX1OR2-NEXT: retq
;
; AVX512-LABEL: concat_fcmp_v16f32_v4f32:
; AVX512: # %bb.0:
; AVX512-NEXT: # kill: def $xmm2 killed $xmm2 def $ymm2
; AVX512-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX512-NEXT: vinsertf128 $1, %xmm3, %ymm2, %ymm2
; AVX512-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX512-NEXT: vinsertf64x4 $1, %ymm2, %zmm0, %zmm0
; AVX512-NEXT: vxorps %xmm1, %xmm1, %xmm1
; AVX512-NEXT: vcmpleps %zmm0, %zmm1, %k0
; AVX512-NEXT: kmovd %k0, %eax
; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%v0 = fcmp oge <4 x float> %a0, zeroinitializer
%v1 = fcmp oge <4 x float> %a1, zeroinitializer
%v2 = fcmp oge <4 x float> %a2, zeroinitializer
%v3 = fcmp oge <4 x float> %a3, zeroinitializer
%v01 = shufflevector <4 x i1> %v0, <4 x i1> %v1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
%v23 = shufflevector <4 x i1> %v2, <4 x i1> %v3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
%v = shufflevector <8 x i1> %v01, <8 x i1> %v23, <16 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>
%r = bitcast <16 x i1> %v to i16
ret i16 %r
}

define i8 @concat_fcmp_v8f64_v4f64(<4 x double> %a0, <4 x double> %a1) {
; SSE-LABEL: concat_fcmp_v8f64_v4f64:
; SSE: # %bb.0:
; SSE-NEXT: xorpd %xmm4, %xmm4
; SSE-NEXT: movapd %xmm1, %xmm5
; SSE-NEXT: cmpneqpd %xmm4, %xmm5
; SSE-NEXT: cmpordpd %xmm4, %xmm1
; SSE-NEXT: andpd %xmm5, %xmm1
; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE-NEXT: movapd %xmm0, %xmm5
; SSE-NEXT: cmpneqpd %xmm4, %xmm5
; SSE-NEXT: cmpordpd %xmm4, %xmm0
; SSE-NEXT: andpd %xmm5, %xmm0
; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT: movapd %xmm3, %xmm1
; SSE-NEXT: cmpneqpd %xmm4, %xmm1
; SSE-NEXT: cmpordpd %xmm4, %xmm3
; SSE-NEXT: andpd %xmm1, %xmm3
; SSE-NEXT: movapd %xmm2, %xmm1
; SSE-NEXT: cmpneqpd %xmm4, %xmm1
; SSE-NEXT: cmpordpd %xmm4, %xmm2
; SSE-NEXT: andpd %xmm1, %xmm2
; SSE-NEXT: packssdw %xmm3, %xmm2
; SSE-NEXT: packssdw %xmm2, %xmm2
; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
; SSE-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
; SSE-NEXT: packsswb %xmm0, %xmm0
; SSE-NEXT: pmovmskb %xmm0, %eax
; SSE-NEXT: # kill: def $al killed $al killed $eax
; SSE-NEXT: retq
;
; AVX1-LABEL: concat_fcmp_v8f64_v4f64:
; AVX1: # %bb.0:
; AVX1-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; AVX1-NEXT: vcmpneq_oqpd %ymm2, %ymm0, %ymm0
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm3
; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm3[0,2]
; AVX1-NEXT: vcmpneq_oqpd %ymm2, %ymm1, %ymm1
; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm1[0,2],xmm2[0,2]
; AVX1-NEXT: vpackssdw %xmm1, %xmm0, %xmm0
; AVX1-NEXT: vpsrlw $8, %xmm0, %xmm0
; AVX1-NEXT: vpackuswb %xmm0, %xmm0, %xmm0
; AVX1-NEXT: vpmovmskb %xmm0, %eax
; AVX1-NEXT: # kill: def $al killed $al killed $eax
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
;
; AVX2-LABEL: concat_fcmp_v8f64_v4f64:
; AVX2: # %bb.0:
; AVX2-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; AVX2-NEXT: vcmpneq_oqpd %ymm2, %ymm0, %ymm0
; AVX2-NEXT: vextractf128 $1, %ymm0, %xmm3
; AVX2-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm3[0,2]
; AVX2-NEXT: vcmpneq_oqpd %ymm2, %ymm1, %ymm1
; AVX2-NEXT: vextractf128 $1, %ymm1, %xmm2
; AVX2-NEXT: vshufps {{.*#+}} xmm1 = xmm1[0,2],xmm2[0,2]
; AVX2-NEXT: vpackssdw %xmm1, %xmm0, %xmm0
; AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u]
; AVX2-NEXT: vpmovmskb %xmm0, %eax
; AVX2-NEXT: # kill: def $al killed $al killed $eax
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
; AVX512-LABEL: concat_fcmp_v8f64_v4f64:
; AVX512: # %bb.0:
; AVX512-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0
; AVX512-NEXT: vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; AVX512-NEXT: vcmpneq_oqpd %zmm1, %zmm0, %k0
; AVX512-NEXT: kmovd %k0, %eax
; AVX512-NEXT: # kill: def $al killed $al killed $eax
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%v0 = fcmp one <4 x double> %a0, zeroinitializer
%v1 = fcmp one <4 x double> %a1, zeroinitializer
%v = shufflevector <4 x i1> %v0, <4 x i1> %v1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
%r = bitcast <8 x i1> %v to i8
ret i8 %r
}

define i16 @concat_fcmp_v16f32_v8f32(<8 x float> %a0, <8 x float> %a1) {
; SSE-LABEL: concat_fcmp_v16f32_v8f32:
; SSE: # %bb.0:
; SSE-NEXT: xorps %xmm4, %xmm4
; SSE-NEXT: cmpleps %xmm4, %xmm1
; SSE-NEXT: cmpleps %xmm4, %xmm0
; SSE-NEXT: packssdw %xmm1, %xmm0
; SSE-NEXT: cmpleps %xmm4, %xmm3
; SSE-NEXT: cmpleps %xmm4, %xmm2
; SSE-NEXT: packssdw %xmm3, %xmm2
; SSE-NEXT: packsswb %xmm2, %xmm0
; SSE-NEXT: pmovmskb %xmm0, %eax
; SSE-NEXT: # kill: def $ax killed $ax killed $eax
; SSE-NEXT: retq
;
; AVX1OR2-LABEL: concat_fcmp_v16f32_v8f32:
; AVX1OR2: # %bb.0:
; AVX1OR2-NEXT: vxorps %xmm2, %xmm2, %xmm2
; AVX1OR2-NEXT: vcmpleps %ymm2, %ymm0, %ymm0
; AVX1OR2-NEXT: vextractf128 $1, %ymm0, %xmm3
; AVX1OR2-NEXT: vpackssdw %xmm3, %xmm0, %xmm0
; AVX1OR2-NEXT: vcmpleps %ymm2, %ymm1, %ymm1
; AVX1OR2-NEXT: vextractf128 $1, %ymm1, %xmm2
; AVX1OR2-NEXT: vpackssdw %xmm2, %xmm1, %xmm1
; AVX1OR2-NEXT: vpacksswb %xmm1, %xmm0, %xmm0
; AVX1OR2-NEXT: vpmovmskb %xmm0, %eax
; AVX1OR2-NEXT: # kill: def $ax killed $ax killed $eax
; AVX1OR2-NEXT: vzeroupper
; AVX1OR2-NEXT: retq
;
; AVX512-LABEL: concat_fcmp_v16f32_v8f32:
; AVX512: # %bb.0:
; AVX512-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0
; AVX512-NEXT: vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
; AVX512-NEXT: vxorps %xmm1, %xmm1, %xmm1
; AVX512-NEXT: vcmpleps %zmm1, %zmm0, %k0
; AVX512-NEXT: kmovd %k0, %eax
; AVX512-NEXT: # kill: def $ax killed $ax killed $eax
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
%v0 = fcmp ole <8 x float> %a0, zeroinitializer
%v1 = fcmp ole <8 x float> %a1, zeroinitializer
%v = shufflevector <8 x i1> %v0, <8 x i1> %v1, <16 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>
%r = bitcast <16 x i1> %v to i16
ret i16 %r
}
Loading