1 ; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
2 ; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
4 declare double @llvm.fma.f64(double, double, double) nounwind readnone
5 declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>) nounwind readnone
6 declare <4 x double> @llvm.fma.v4f64(<4 x double>, <4 x double>, <4 x double>) nounwind readnone
9 ; FUNC-LABEL: {{^}}fma_f64:
10 ; SI: v_fma_f64 {{v\[[0-9]+:[0-9]+\], v\[[0-9]+:[0-9]+\], v\[[0-9]+:[0-9]+\], v\[[0-9]+:[0-9]+\]}}
11 define amdgpu_kernel void @fma_f64(double addrspace(1)* %out, double addrspace(1)* %in1,
12 double addrspace(1)* %in2, double addrspace(1)* %in3) {
13 %r0 = load double, double addrspace(1)* %in1
14 %r1 = load double, double addrspace(1)* %in2
15 %r2 = load double, double addrspace(1)* %in3
16 %r3 = tail call double @llvm.fma.f64(double %r0, double %r1, double %r2)
17 store double %r3, double addrspace(1)* %out
21 ; FUNC-LABEL: {{^}}fma_v2f64:
24 define amdgpu_kernel void @fma_v2f64(<2 x double> addrspace(1)* %out, <2 x double> addrspace(1)* %in1,
25 <2 x double> addrspace(1)* %in2, <2 x double> addrspace(1)* %in3) {
26 %r0 = load <2 x double>, <2 x double> addrspace(1)* %in1
27 %r1 = load <2 x double>, <2 x double> addrspace(1)* %in2
28 %r2 = load <2 x double>, <2 x double> addrspace(1)* %in3
29 %r3 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %r0, <2 x double> %r1, <2 x double> %r2)
30 store <2 x double> %r3, <2 x double> addrspace(1)* %out
34 ; FUNC-LABEL: {{^}}fma_v4f64:
39 define amdgpu_kernel void @fma_v4f64(<4 x double> addrspace(1)* %out, <4 x double> addrspace(1)* %in1,
40 <4 x double> addrspace(1)* %in2, <4 x double> addrspace(1)* %in3) {
41 %r0 = load <4 x double>, <4 x double> addrspace(1)* %in1
42 %r1 = load <4 x double>, <4 x double> addrspace(1)* %in2
43 %r2 = load <4 x double>, <4 x double> addrspace(1)* %in3
44 %r3 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %r0, <4 x double> %r1, <4 x double> %r2)
45 store <4 x double> %r3, <4 x double> addrspace(1)* %out