summaryrefslogtreecommitdiff
path: root/test/CodeGen/X86/avx-cmp.ll
diff options
context:
space:
mode:
Diffstat (limited to 'test/CodeGen/X86/avx-cmp.ll')
-rw-r--r--test/CodeGen/X86/avx-cmp.ll197
1 files changed, 133 insertions, 64 deletions
diff --git a/test/CodeGen/X86/avx-cmp.ll b/test/CodeGen/X86/avx-cmp.ll
index a050d6abe56f..963878b0f563 100644
--- a/test/CodeGen/X86/avx-cmp.ll
+++ b/test/CodeGen/X86/avx-cmp.ll
@@ -1,25 +1,59 @@
-; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=corei7-avx -mattr=+avx | FileCheck %s
-
-; CHECK: vcmpltps %ymm
-; CHECK-NOT: vucomiss
-define <8 x i32> @cmp00(<8 x float> %a, <8 x float> %b) nounwind readnone {
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s
+
+define <8 x i32> @cmp00(<8 x float> %a, <8 x float> %b) nounwind {
+; CHECK-LABEL: cmp00:
+; CHECK: # BB#0:
+; CHECK-NEXT: vcmpltps %ymm1, %ymm0, %ymm0
+; CHECK-NEXT: retq
%bincmp = fcmp olt <8 x float> %a, %b
%s = sext <8 x i1> %bincmp to <8 x i32>
ret <8 x i32> %s
}
-; CHECK: vcmpltpd %ymm
-; CHECK-NOT: vucomisd
-define <4 x i64> @cmp01(<4 x double> %a, <4 x double> %b) nounwind readnone {
+define <4 x i64> @cmp01(<4 x double> %a, <4 x double> %b) nounwind {
+; CHECK-LABEL: cmp01:
+; CHECK: # BB#0:
+; CHECK-NEXT: vcmpltpd %ymm1, %ymm0, %ymm0
+; CHECK-NEXT: retq
%bincmp = fcmp olt <4 x double> %a, %b
%s = sext <4 x i1> %bincmp to <4 x i64>
ret <4 x i64> %s
}
-declare void @scale() nounwind uwtable
-
-; CHECK: vucomisd
-define void @render() nounwind uwtable {
+declare void @scale() nounwind
+
+define void @render() nounwind {
+; CHECK-LABEL: render:
+; CHECK: # BB#0: # %entry
+; CHECK-NEXT: pushq %rbx
+; CHECK-NEXT: xorl %eax, %eax
+; CHECK-NEXT: testb %al, %al
+; CHECK-NEXT: jne .LBB2_6
+; CHECK-NEXT: # BB#1: # %for.cond5.preheader
+; CHECK-NEXT: xorl %ebx, %ebx
+; CHECK-NEXT: jmp .LBB2_2
+; CHECK-NEXT: .p2align 4, 0x90
+; CHECK-NEXT: .LBB2_5: # %if.then
+; CHECK-NEXT: # in Loop: Header=BB2_2 Depth=1
+; CHECK-NEXT: callq scale
+; CHECK-NEXT: .LBB2_2: # %for.cond5
+; CHECK-NEXT: # =>This Inner Loop Header: Depth=1
+; CHECK-NEXT: testb %bl, %bl
+; CHECK-NEXT: jne .LBB2_2
+; CHECK-NEXT: # BB#3: # %for.cond5
+; CHECK-NEXT: # in Loop: Header=BB2_2 Depth=1
+; CHECK-NEXT: testb %bl, %bl
+; CHECK-NEXT: je .LBB2_2
+; CHECK-NEXT: # BB#4: # %for.body33
+; CHECK-NEXT: # in Loop: Header=BB2_2 Depth=1
+; CHECK-NEXT: vucomisd {{\.LCPI.*}}, %xmm0
+; CHECK-NEXT: jne .LBB2_5
+; CHECK-NEXT: jp .LBB2_5
+; CHECK-NEXT: jmp .LBB2_2
+; CHECK-NEXT: .LBB2_6: # %for.end52
+; CHECK-NEXT: popq %rbx
+; CHECK-NEXT: retq
entry:
br i1 undef, label %for.cond5, label %for.end52
@@ -42,89 +76,113 @@ for.end52:
ret void
}
-; CHECK: vextractf128 $1
-; CHECK: vextractf128 $1
-; CHECK-NEXT: vpcmpgtd %xmm
-; CHECK-NEXT: vpcmpgtd %xmm
-; CHECK-NEXT: vinsertf128 $1
-define <8 x i32> @int256-cmp(<8 x i32> %i, <8 x i32> %j) nounwind readnone {
+define <8 x i32> @int256_cmp(<8 x i32> %i, <8 x i32> %j) nounwind {
+; CHECK-LABEL: int256_cmp:
+; CHECK: # BB#0:
+; CHECK-NEXT: vextractf128 $1, %ymm0, %xmm2
+; CHECK-NEXT: vextractf128 $1, %ymm1, %xmm3
+; CHECK-NEXT: vpcmpgtd %xmm2, %xmm3, %xmm2
+; CHECK-NEXT: vpcmpgtd %xmm0, %xmm1, %xmm0
+; CHECK-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; CHECK-NEXT: retq
%bincmp = icmp slt <8 x i32> %i, %j
%x = sext <8 x i1> %bincmp to <8 x i32>
ret <8 x i32> %x
}
-; CHECK: vextractf128 $1
-; CHECK: vextractf128 $1
-; CHECK-NEXT: vpcmpgtq %xmm
-; CHECK-NEXT: vpcmpgtq %xmm
-; CHECK-NEXT: vinsertf128 $1
-define <4 x i64> @v4i64-cmp(<4 x i64> %i, <4 x i64> %j) nounwind readnone {
+define <4 x i64> @v4i64_cmp(<4 x i64> %i, <4 x i64> %j) nounwind {
+; CHECK-LABEL: v4i64_cmp:
+; CHECK: # BB#0:
+; CHECK-NEXT: vextractf128 $1, %ymm0, %xmm2
+; CHECK-NEXT: vextractf128 $1, %ymm1, %xmm3
+; CHECK-NEXT: vpcmpgtq %xmm2, %xmm3, %xmm2
+; CHECK-NEXT: vpcmpgtq %xmm0, %xmm1, %xmm0
+; CHECK-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; CHECK-NEXT: retq
%bincmp = icmp slt <4 x i64> %i, %j
%x = sext <4 x i1> %bincmp to <4 x i64>
ret <4 x i64> %x
}
-; CHECK: vextractf128 $1
-; CHECK: vextractf128 $1
-; CHECK-NEXT: vpcmpgtw %xmm
-; CHECK-NEXT: vpcmpgtw %xmm
-; CHECK-NEXT: vinsertf128 $1
-define <16 x i16> @v16i16-cmp(<16 x i16> %i, <16 x i16> %j) nounwind readnone {
+define <16 x i16> @v16i16_cmp(<16 x i16> %i, <16 x i16> %j) nounwind {
+; CHECK-LABEL: v16i16_cmp:
+; CHECK: # BB#0:
+; CHECK-NEXT: vextractf128 $1, %ymm0, %xmm2
+; CHECK-NEXT: vextractf128 $1, %ymm1, %xmm3
+; CHECK-NEXT: vpcmpgtw %xmm2, %xmm3, %xmm2
+; CHECK-NEXT: vpcmpgtw %xmm0, %xmm1, %xmm0
+; CHECK-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; CHECK-NEXT: retq
%bincmp = icmp slt <16 x i16> %i, %j
%x = sext <16 x i1> %bincmp to <16 x i16>
ret <16 x i16> %x
}
-; CHECK: vextractf128 $1
-; CHECK: vextractf128 $1
-; CHECK-NEXT: vpcmpgtb %xmm
-; CHECK-NEXT: vpcmpgtb %xmm
-; CHECK-NEXT: vinsertf128 $1
-define <32 x i8> @v32i8-cmp(<32 x i8> %i, <32 x i8> %j) nounwind readnone {
+define <32 x i8> @v32i8_cmp(<32 x i8> %i, <32 x i8> %j) nounwind {
+; CHECK-LABEL: v32i8_cmp:
+; CHECK: # BB#0:
+; CHECK-NEXT: vextractf128 $1, %ymm0, %xmm2
+; CHECK-NEXT: vextractf128 $1, %ymm1, %xmm3
+; CHECK-NEXT: vpcmpgtb %xmm2, %xmm3, %xmm2
+; CHECK-NEXT: vpcmpgtb %xmm0, %xmm1, %xmm0
+; CHECK-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; CHECK-NEXT: retq
%bincmp = icmp slt <32 x i8> %i, %j
%x = sext <32 x i1> %bincmp to <32 x i8>
ret <32 x i8> %x
}
-; CHECK: vextractf128 $1
-; CHECK: vextractf128 $1
-; CHECK-NEXT: vpcmpeqd %xmm
-; CHECK-NEXT: vpcmpeqd %xmm
-; CHECK-NEXT: vinsertf128 $1
-define <8 x i32> @int256-cmpeq(<8 x i32> %i, <8 x i32> %j) nounwind readnone {
+define <8 x i32> @int256_cmpeq(<8 x i32> %i, <8 x i32> %j) nounwind {
+; CHECK-LABEL: int256_cmpeq:
+; CHECK: # BB#0:
+; CHECK-NEXT: vextractf128 $1, %ymm1, %xmm2
+; CHECK-NEXT: vextractf128 $1, %ymm0, %xmm3
+; CHECK-NEXT: vpcmpeqd %xmm2, %xmm3, %xmm2
+; CHECK-NEXT: vpcmpeqd %xmm1, %xmm0, %xmm0
+; CHECK-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; CHECK-NEXT: retq
%bincmp = icmp eq <8 x i32> %i, %j
%x = sext <8 x i1> %bincmp to <8 x i32>
ret <8 x i32> %x
}
-; CHECK: vextractf128 $1
-; CHECK: vextractf128 $1
-; CHECK-NEXT: vpcmpeqq %xmm
-; CHECK-NEXT: vpcmpeqq %xmm
-; CHECK-NEXT: vinsertf128 $1
-define <4 x i64> @v4i64-cmpeq(<4 x i64> %i, <4 x i64> %j) nounwind readnone {
+define <4 x i64> @v4i64_cmpeq(<4 x i64> %i, <4 x i64> %j) nounwind {
+; CHECK-LABEL: v4i64_cmpeq:
+; CHECK: # BB#0:
+; CHECK-NEXT: vextractf128 $1, %ymm1, %xmm2
+; CHECK-NEXT: vextractf128 $1, %ymm0, %xmm3
+; CHECK-NEXT: vpcmpeqq %xmm2, %xmm3, %xmm2
+; CHECK-NEXT: vpcmpeqq %xmm1, %xmm0, %xmm0
+; CHECK-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; CHECK-NEXT: retq
%bincmp = icmp eq <4 x i64> %i, %j
%x = sext <4 x i1> %bincmp to <4 x i64>
ret <4 x i64> %x
}
-; CHECK: vextractf128 $1
-; CHECK: vextractf128 $1
-; CHECK-NEXT: vpcmpeqw %xmm
-; CHECK-NEXT: vpcmpeqw %xmm
-; CHECK-NEXT: vinsertf128 $1
-define <16 x i16> @v16i16-cmpeq(<16 x i16> %i, <16 x i16> %j) nounwind readnone {
+define <16 x i16> @v16i16_cmpeq(<16 x i16> %i, <16 x i16> %j) nounwind {
+; CHECK-LABEL: v16i16_cmpeq:
+; CHECK: # BB#0:
+; CHECK-NEXT: vextractf128 $1, %ymm1, %xmm2
+; CHECK-NEXT: vextractf128 $1, %ymm0, %xmm3
+; CHECK-NEXT: vpcmpeqw %xmm2, %xmm3, %xmm2
+; CHECK-NEXT: vpcmpeqw %xmm1, %xmm0, %xmm0
+; CHECK-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; CHECK-NEXT: retq
%bincmp = icmp eq <16 x i16> %i, %j
%x = sext <16 x i1> %bincmp to <16 x i16>
ret <16 x i16> %x
}
-; CHECK: vextractf128 $1
-; CHECK: vextractf128 $1
-; CHECK-NEXT: vpcmpeqb %xmm
-; CHECK-NEXT: vpcmpeqb %xmm
-; CHECK-NEXT: vinsertf128 $1
-define <32 x i8> @v32i8-cmpeq(<32 x i8> %i, <32 x i8> %j) nounwind readnone {
+define <32 x i8> @v32i8_cmpeq(<32 x i8> %i, <32 x i8> %j) nounwind {
+; CHECK-LABEL: v32i8_cmpeq:
+; CHECK: # BB#0:
+; CHECK-NEXT: vextractf128 $1, %ymm1, %xmm2
+; CHECK-NEXT: vextractf128 $1, %ymm0, %xmm3
+; CHECK-NEXT: vpcmpeqb %xmm2, %xmm3, %xmm2
+; CHECK-NEXT: vpcmpeqb %xmm1, %xmm0, %xmm0
+; CHECK-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; CHECK-NEXT: retq
%bincmp = icmp eq <32 x i8> %i, %j
%x = sext <32 x i1> %bincmp to <32 x i8>
ret <32 x i8> %x
@@ -132,17 +190,28 @@ define <32 x i8> @v32i8-cmpeq(<32 x i8> %i, <32 x i8> %j) nounwind readnone {
;; Scalar comparison
-; CHECK: scalarcmpA
-; CHECK: vcmpeqsd
define i32 @scalarcmpA() uwtable ssp {
+; CHECK-LABEL: scalarcmpA:
+; CHECK: # BB#0:
+; CHECK-NEXT: vxorpd %xmm0, %xmm0, %xmm0
+; CHECK-NEXT: vcmpeqsd %xmm0, %xmm0, %xmm0
+; CHECK-NEXT: vmovq %xmm0, %rax
+; CHECK-NEXT: andl $1, %eax
+; CHECK-NEXT: # kill: %EAX<def> %EAX<kill> %RAX<kill>
+; CHECK-NEXT: retq
%cmp29 = fcmp oeq double undef, 0.000000e+00
%res = zext i1 %cmp29 to i32
ret i32 %res
}
-; CHECK: scalarcmpB
-; CHECK: vcmpeqss
define i32 @scalarcmpB() uwtable ssp {
+; CHECK-LABEL: scalarcmpB:
+; CHECK: # BB#0:
+; CHECK-NEXT: vxorps %xmm0, %xmm0, %xmm0
+; CHECK-NEXT: vcmpeqss %xmm0, %xmm0, %xmm0
+; CHECK-NEXT: vmovd %xmm0, %eax
+; CHECK-NEXT: andl $1, %eax
+; CHECK-NEXT: retq
%cmp29 = fcmp oeq float undef, 0.000000e+00
%res = zext i1 %cmp29 to i32
ret i32 %res
generated by cgit v1.2.3 (git 2.25.1) at 2025-08-01 19:14:33 +0000