1 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32-n8:16:32-S128" 2 target triple = "i686-unknown-linux" 3 4 declare <4 x float> @llvm.x86.sse.min.ps(<4 x float>, <4 x float>) 5 declare <4 x float> @llvm.x86.sse.max.ps(<4 x float>, <4 x float>) 6 declare <4 x float> @llvm.x86.sse.min.ss(<4 x float>, <4 x float>) 7 declare <4 x float> @llvm.x86.sse.max.ss(<4 x float>, <4 x float>) 8 9 declare float @llvm.sqrt.f32(float) nounwind readnone 10 declare <2 x float> @llvm.sqrt.v2f32(<2 x float>) nounwind readnone 11 declare <3 x float> @llvm.sqrt.v3f32(<3 x float>) nounwind readnone 12 declare <4 x float> @llvm.sqrt.v4f32(<4 x float>) nounwind readnone 13 14 declare float @llvm.exp.f32(float) nounwind readonly 15 declare float @llvm.pow.f32(float, float) nounwind readonly 16 17 define <4 x float> @_Z5clampDv4_fS_S_(<4 x float> %in, <4 x float> %low, <4 x float> %high) nounwind readnone alwaysinline { 18 %1 = tail call <4 x float> @llvm.x86.sse.min.ps(<4 x float> %in, <4 x float> %high) nounwind readnone 19 %2 = tail call <4 x float> @llvm.x86.sse.max.ps(<4 x float> %1, <4 x float> %low) nounwind readnone 20 ret <4 x float> %2 21 } 22 23 define <3 x float> @_Z5clampDv3_fS_S_(<3 x float> %in, <3 x float> %low, <3 x float> %high) nounwind readnone alwaysinline { 24 %1 = shufflevector <3 x float> %in, <3 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3> 25 %2 = shufflevector <3 x float> %low, <3 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3> 26 %3 = shufflevector <3 x float> %high, <3 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3> 27 %4 = tail call <4 x float> @_Z5clampDv4_fS_S_(<4 x float> %1, <4 x float> %2, <4 x float> %3) nounwind readnone 28 %5 = shufflevector <4 x float> %4, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2> 29 ret <3 x float> %5 30 } 31 32 define <2 x float> @_Z5clampDv2_fS_S_(<2 x float> %in, <2 x float> %low, <2 x float> %high) nounwind readnone alwaysinline { 33 %1 = shufflevector <2 x float> %in, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3> 34 %2 = shufflevector <2 x float> %low, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3> 35 %3 = shufflevector <2 x float> %high, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3> 36 %4 = tail call <4 x float> @_Z5clampDv4_fS_S_(<4 x float> %1, <4 x float> %2, <4 x float> %3) nounwind readnone 37 %5 = shufflevector <4 x float> %4, <4 x float> undef, <2 x i32> <i32 0, i32 1> 38 ret <2 x float> %5 39 } 40 41 define float @_Z5clampfff(float %in, float %low, float %high) nounwind readnone alwaysinline { 42 %1 = insertelement <4 x float> undef, float %in, i32 0 43 %2 = insertelement <4 x float> undef, float %low, i32 0 44 %3 = insertelement <4 x float> undef, float %high, i32 0 45 %4 = tail call <4 x float> @llvm.x86.sse.min.ss(<4 x float> %1, <4 x float> %3) nounwind readnone 46 %5 = tail call <4 x float> @llvm.x86.sse.max.ss(<4 x float> %4, <4 x float> %2) nounwind readnone 47 %6 = extractelement <4 x float> %5, i32 0 48 ret float %6 49 } 50 51 define <4 x float> @_Z5clampDv4_fff(<4 x float> %in, float %low, float %high) nounwind readonly { 52 %1 = insertelement <4 x float> undef, float %low, i32 0 53 %2 = insertelement <4 x float> %1, float %low, i32 1 54 %3 = insertelement <4 x float> %2, float %low, i32 2 55 %4 = insertelement <4 x float> %3, float %low, i32 3 56 %5 = insertelement <4 x float> undef, float %high, i32 0 57 %6 = insertelement <4 x float> %5, float %high, i32 1 58 %7 = insertelement <4 x float> %6, float %high, i32 2 59 %8 = insertelement <4 x float> %7, float %high, i32 3 60 %9 = tail call <4 x float> @_Z5clampDv4_fS_S_(<4 x float> %in, <4 x float> %4, <4 x float> %8) nounwind readnone 61 ret <4 x float> %9 62 } 63 64 define <3 x float> @_Z5clampDv3_fff(<3 x float> %in, float %low, float %high) nounwind readonly { 65 %1 = insertelement <3 x float> undef, float %low, i32 0 66 %2 = insertelement <3 x float> %1, float %low, i32 1 67 %3 = insertelement <3 x float> %2, float %low, i32 2 68 %4 = insertelement <3 x float> undef, float %high, i32 0 69 %5 = insertelement <3 x float> %4, float %high, i32 1 70 %6 = insertelement <3 x float> %5, float %high, i32 2 71 %7 = tail call <3 x float> @_Z5clampDv3_fS_S_(<3 x float> %in, <3 x float> %3, <3 x float> %6) nounwind readnone 72 ret <3 x float> %7 73 } 74 75 define <2 x float> @_Z5clampDv2_fff(<2 x float> %in, float %low, float %high) nounwind readonly { 76 %1 = insertelement <2 x float> undef, float %low, i32 0 77 %2 = insertelement <2 x float> %1, float %low, i32 1 78 %3 = insertelement <2 x float> undef, float %high, i32 0 79 %4 = insertelement <2 x float> %3, float %high, i32 1 80 %5 = tail call <2 x float> @_Z5clampDv2_fS_S_(<2 x float> %in, <2 x float> %2, <2 x float> %4) nounwind readnone 81 ret <2 x float> %5 82 } 83 84 define float @_Z4sqrtf(float %in) nounwind readnone alwaysinline { 85 %1 = tail call float @llvm.sqrt.f32(float %in) nounwind readnone 86 ret float %1 87 } 88 89 define <2 x float> @_Z4sqrtDv2_f(<2 x float> %in) nounwind readnone alwaysinline { 90 %1 = tail call <2 x float> @llvm.sqrt.v2f32(<2 x float> %in) nounwind readnone 91 ret <2 x float> %1 92 } 93 94 define <3 x float> @_Z4sqrtDv3_f(<3 x float> %in) nounwind readnone alwaysinline { 95 %1 = tail call <3 x float> @llvm.sqrt.v3f32(<3 x float> %in) nounwind readnone 96 ret <3 x float> %1 97 } 98 99 define <4 x float> @_Z4sqrtDv4_f(<4 x float> %in) nounwind readnone alwaysinline { 100 %1 = tail call <4 x float> @llvm.sqrt.v4f32(<4 x float> %in) nounwind readnone 101 ret <4 x float> %1 102 } 103
