1; RUN: llc -march=r600 -mcpu=SI -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s 2 3declare float @llvm.fma.f32(float, float, float) nounwind readnone 4declare <2 x float> @llvm.fma.v2f32(<2 x float>, <2 x float>, <2 x float>) nounwind readnone 5declare <4 x float> @llvm.fma.v4f32(<4 x float>, <4 x float>, <4 x float>) nounwind readnone 6 7declare double @llvm.fma.f64(double, double, double) nounwind readnone 8declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>) nounwind readnone 9declare <4 x double> @llvm.fma.v4f64(<4 x double>, <4 x double>, <4 x double>) nounwind readnone 10 11; FUNC-LABEL: @fma_f32 12; SI: V_FMA_F32 {{v[0-9]+, v[0-9]+, v[0-9]+, v[0-9]+}} 13define void @fma_f32(float addrspace(1)* %out, float addrspace(1)* %in1, 14 float addrspace(1)* %in2, float addrspace(1)* %in3) { 15 %r0 = load float addrspace(1)* %in1 16 %r1 = load float addrspace(1)* %in2 17 %r2 = load float addrspace(1)* %in3 18 %r3 = tail call float @llvm.fma.f32(float %r0, float %r1, float %r2) 19 store float %r3, float addrspace(1)* %out 20 ret void 21} 22 23; FUNC-LABEL: @fma_v2f32 24; SI: V_FMA_F32 25; SI: V_FMA_F32 26define void @fma_v2f32(<2 x float> addrspace(1)* %out, <2 x float> addrspace(1)* %in1, 27 <2 x float> addrspace(1)* %in2, <2 x float> addrspace(1)* %in3) { 28 %r0 = load <2 x float> addrspace(1)* %in1 29 %r1 = load <2 x float> addrspace(1)* %in2 30 %r2 = load <2 x float> addrspace(1)* %in3 31 %r3 = tail call <2 x float> @llvm.fma.v2f32(<2 x float> %r0, <2 x float> %r1, <2 x float> %r2) 32 store <2 x float> %r3, <2 x float> addrspace(1)* %out 33 ret void 34} 35 36; FUNC-LABEL: @fma_v4f32 37; SI: V_FMA_F32 38; SI: V_FMA_F32 39; SI: V_FMA_F32 40; SI: V_FMA_F32 41define void @fma_v4f32(<4 x float> addrspace(1)* %out, <4 x float> addrspace(1)* %in1, 42 <4 x float> addrspace(1)* %in2, <4 x float> addrspace(1)* %in3) { 43 %r0 = load <4 x float> addrspace(1)* %in1 44 %r1 = load <4 x float> addrspace(1)* %in2 45 %r2 = load <4 x float> addrspace(1)* %in3 46 %r3 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %r0, <4 x float> %r1, <4 x float> %r2) 47 store <4 x float> %r3, <4 x float> addrspace(1)* %out 48 ret void 49} 50 51; FUNC-LABEL: @fma_f64 52; SI: V_FMA_F64 {{v\[[0-9]+:[0-9]+\], v\[[0-9]+:[0-9]+\], v\[[0-9]+:[0-9]+\], v\[[0-9]+:[0-9]+\]}} 53define void @fma_f64(double addrspace(1)* %out, double addrspace(1)* %in1, 54 double addrspace(1)* %in2, double addrspace(1)* %in3) { 55 %r0 = load double addrspace(1)* %in1 56 %r1 = load double addrspace(1)* %in2 57 %r2 = load double addrspace(1)* %in3 58 %r3 = tail call double @llvm.fma.f64(double %r0, double %r1, double %r2) 59 store double %r3, double addrspace(1)* %out 60 ret void 61} 62 63; FUNC-LABEL: @fma_v2f64 64; SI: V_FMA_F64 65; SI: V_FMA_F64 66define void @fma_v2f64(<2 x double> addrspace(1)* %out, <2 x double> addrspace(1)* %in1, 67 <2 x double> addrspace(1)* %in2, <2 x double> addrspace(1)* %in3) { 68 %r0 = load <2 x double> addrspace(1)* %in1 69 %r1 = load <2 x double> addrspace(1)* %in2 70 %r2 = load <2 x double> addrspace(1)* %in3 71 %r3 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %r0, <2 x double> %r1, <2 x double> %r2) 72 store <2 x double> %r3, <2 x double> addrspace(1)* %out 73 ret void 74} 75 76; FUNC-LABEL: @fma_v4f64 77; SI: V_FMA_F64 78; SI: V_FMA_F64 79; SI: V_FMA_F64 80; SI: V_FMA_F64 81define void @fma_v4f64(<4 x double> addrspace(1)* %out, <4 x double> addrspace(1)* %in1, 82 <4 x double> addrspace(1)* %in2, <4 x double> addrspace(1)* %in3) { 83 %r0 = load <4 x double> addrspace(1)* %in1 84 %r1 = load <4 x double> addrspace(1)* %in2 85 %r2 = load <4 x double> addrspace(1)* %in3 86 %r3 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %r0, <4 x double> %r1, <4 x double> %r2) 87 store <4 x double> %r3, <4 x double> addrspace(1)* %out 88 ret void 89} 90