1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py 2 ; RUN: llc < %s -mtriple=i686-apple-darwin9 -mattr=+sse,+sse2,+sse4.1 | FileCheck %s 3 4 target datalayout = "e-p:32:32" 5 6 define i16 @test1(float %f) nounwind { 7 ; CHECK-LABEL: test1: 8 ; CHECK: ## BB#0: 9 ; CHECK-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero 10 ; CHECK-NEXT: xorps %xmm1, %xmm1 11 ; CHECK-NEXT: subss LCPI0_0, %xmm0 12 ; CHECK-NEXT: mulss LCPI0_1, %xmm0 13 ; CHECK-NEXT: minss LCPI0_2, %xmm0 14 ; CHECK-NEXT: maxss %xmm1, %xmm0 15 ; CHECK-NEXT: cvttss2si %xmm0, %eax 16 ; CHECK-NEXT: ## kill: %AX<def> %AX<kill> %EAX<kill> 17 ; CHECK-NEXT: retl 18 ; 19 %tmp = insertelement <4 x float> undef, float %f, i32 0 ; <<4 x float>> [#uses=1] 20 %tmp10 = insertelement <4 x float> %tmp, float 0.000000e+00, i32 1 ; <<4 x float>> [#uses=1] 21 %tmp11 = insertelement <4 x float> %tmp10, float 0.000000e+00, i32 2 ; <<4 x float>> [#uses=1] 22 %tmp12 = insertelement <4 x float> %tmp11, float 0.000000e+00, i32 3 ; <<4 x float>> [#uses=1] 23 %tmp28 = tail call <4 x float> @llvm.x86.sse.sub.ss( <4 x float> %tmp12, <4 x float> < float 1.000000e+00, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00 > ) ; <<4 x float>> [#uses=1] 24 %tmp37 = tail call <4 x float> @llvm.x86.sse.mul.ss( <4 x float> %tmp28, <4 x float> < float 5.000000e-01, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00 > ) ; <<4 x float>> [#uses=1] 25 %tmp48 = tail call <4 x float> @llvm.x86.sse.min.ss( <4 x float> %tmp37, <4 x float> < float 6.553500e+04, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00 > ) ; <<4 x float>> [#uses=1] 26 %tmp59 = tail call <4 x float> @llvm.x86.sse.max.ss( <4 x float> %tmp48, <4 x float> zeroinitializer ) ; <<4 x float>> [#uses=1] 27 %tmp.upgrd.1 = tail call i32 @llvm.x86.sse.cvttss2si( <4 x float> %tmp59 ) ; <i32> [#uses=1] 28 %tmp69 = trunc i32 %tmp.upgrd.1 to i16 ; <i16> [#uses=1] 29 ret i16 %tmp69 30 } 31 32 define i16 @test2(float %f) nounwind { 33 ; CHECK-LABEL: test2: 34 ; CHECK: ## BB#0: 35 ; CHECK-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero 36 ; CHECK-NEXT: addss LCPI1_0, %xmm0 37 ; CHECK-NEXT: mulss LCPI1_1, %xmm0 38 ; CHECK-NEXT: minss LCPI1_2, %xmm0 39 ; CHECK-NEXT: xorps %xmm1, %xmm1 40 ; CHECK-NEXT: maxss %xmm1, %xmm0 41 ; CHECK-NEXT: cvttss2si %xmm0, %eax 42 ; CHECK-NEXT: ## kill: %AX<def> %AX<kill> %EAX<kill> 43 ; CHECK-NEXT: retl 44 ; 45 %tmp28 = fsub float %f, 1.000000e+00 ; <float> [#uses=1] 46 %tmp37 = fmul float %tmp28, 5.000000e-01 ; <float> [#uses=1] 47 %tmp375 = insertelement <4 x float> undef, float %tmp37, i32 0 ; <<4 x float>> [#uses=1] 48 %tmp48 = tail call <4 x float> @llvm.x86.sse.min.ss( <4 x float> %tmp375, <4 x float> < float 6.553500e+04, float undef, float undef, float undef > ) ; <<4 x float>> [#uses=1] 49 %tmp59 = tail call <4 x float> @llvm.x86.sse.max.ss( <4 x float> %tmp48, <4 x float> < float 0.000000e+00, float undef, float undef, float undef > ) ; <<4 x float>> [#uses=1] 50 %tmp = tail call i32 @llvm.x86.sse.cvttss2si( <4 x float> %tmp59 ) ; <i32> [#uses=1] 51 %tmp69 = trunc i32 %tmp to i16 ; <i16> [#uses=1] 52 ret i16 %tmp69 53 } 54 55 declare <4 x float> @llvm.x86.sse.sub.ss(<4 x float>, <4 x float>) 56 57 declare <4 x float> @llvm.x86.sse.mul.ss(<4 x float>, <4 x float>) 58 59 declare <4 x float> @llvm.x86.sse.min.ss(<4 x float>, <4 x float>) 60 61 declare <4 x float> @llvm.x86.sse.max.ss(<4 x float>, <4 x float>) 62 63 declare i32 @llvm.x86.sse.cvttss2si(<4 x float>) 64 65 declare <4 x float> @llvm.x86.sse41.round.ss(<4 x float>, <4 x float>, i32) 66 67 declare <4 x float> @f() 68 69 define <4 x float> @test3(<4 x float> %A, float *%b, i32 %C) nounwind { 70 ; CHECK-LABEL: test3: 71 ; CHECK: ## BB#0: 72 ; CHECK-NEXT: movl {{[0-9]+}}(%esp), %eax 73 ; CHECK-NEXT: roundss $4, (%eax), %xmm0 74 ; CHECK-NEXT: retl 75 ; 76 %a = load float , float *%b 77 %B = insertelement <4 x float> undef, float %a, i32 0 78 %X = call <4 x float> @llvm.x86.sse41.round.ss(<4 x float> %A, <4 x float> %B, i32 4) 79 ret <4 x float> %X 80 } 81 82 define <4 x float> @test4(<4 x float> %A, float *%b, i32 %C) nounwind { 83 ; CHECK-LABEL: test4: 84 ; CHECK: ## BB#0: 85 ; CHECK-NEXT: subl $28, %esp 86 ; CHECK-NEXT: movl {{[0-9]+}}(%esp), %eax 87 ; CHECK-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero 88 ; CHECK-NEXT: movaps %xmm0, (%esp) ## 16-byte Spill 89 ; CHECK-NEXT: calll _f 90 ; CHECK-NEXT: movaps (%esp), %xmm1 ## 16-byte Reload 91 ; CHECK-NEXT: roundss $4, %xmm1, %xmm0 92 ; CHECK-NEXT: addl $28, %esp 93 ; CHECK-NEXT: retl 94 ; 95 %a = load float , float *%b 96 %B = insertelement <4 x float> undef, float %a, i32 0 97 %q = call <4 x float> @f() 98 %X = call <4 x float> @llvm.x86.sse41.round.ss(<4 x float> %q, <4 x float> %B, i32 4) 99 ret <4 x float> %X 100 } 101 102 ; PR13576 103 define <2 x double> @test5() nounwind uwtable readnone noinline { 104 ; CHECK-LABEL: test5: 105 ; CHECK: ## BB#0: ## %entry 106 ; CHECK-NEXT: movaps {{.*#+}} xmm0 = [4.569870e+02,1.233210e+02] 107 ; CHECK-NEXT: movl $128, %eax 108 ; CHECK-NEXT: cvtsi2sdl %eax, %xmm0 109 ; CHECK-NEXT: retl 110 ; 111 entry: 112 %0 = tail call <2 x double> @llvm.x86.sse2.cvtsi2sd(<2 x double> <double 113 4.569870e+02, double 1.233210e+02>, i32 128) nounwind readnone 114 ret <2 x double> %0 115 } 116 117 declare <2 x double> @llvm.x86.sse2.cvtsi2sd(<2 x double>, i32) nounwind readnone 118
