1 ; RUN: llc -march=amdgcn -mcpu=SI -enable-misched < %s | FileCheck -check-prefix=SI -check-prefix=GCN -check-prefix=FUNC %s
2 ; RUN: llc -march=amdgcn -mcpu=bonaire -enable-misched < %s | FileCheck -check-prefix=CI -check-prefix=GCN -check-prefix=FUNC %s
3 ; RUN: llc -march=amdgcn -mcpu=tonga -enable-misched < %s | FileCheck -check-prefix=CI -check-prefix=GCN -check-prefix=FUNC %s
5 ; FUNC-LABEL: {{^}}frem_f32:
6 ; GCN-DAG: buffer_load_dword [[X:v[0-9]+]], {{.*$}}
7 ; GCN-DAG: buffer_load_dword [[Y:v[0-9]+]], {{.*}} offset:16
13 ; GCN: v_trunc_f32_e32
16 define void @frem_f32(float addrspace(1)* %out, float addrspace(1)* %in1,
17 float addrspace(1)* %in2) #0 {
18 %gep2 = getelementptr float addrspace(1)* %in2, i32 4
19 %r0 = load float addrspace(1)* %in1, align 4
20 %r1 = load float addrspace(1)* %gep2, align 4
21 %r2 = frem float %r0, %r1
22 store float %r2, float addrspace(1)* %out, align 4
26 ; FUNC-LABEL: {{^}}unsafe_frem_f32:
27 ; GCN: buffer_load_dword [[Y:v[0-9]+]], {{.*}} offset:16
28 ; GCN: buffer_load_dword [[X:v[0-9]+]], {{.*}}
29 ; GCN: v_rcp_f32_e32 [[INVY:v[0-9]+]], [[Y]]
30 ; GCN: v_mul_f32_e32 [[DIV:v[0-9]+]], [[INVY]], [[X]]
31 ; GCN: v_trunc_f32_e32 [[TRUNC:v[0-9]+]], [[DIV]]
32 ; GCN: v_mad_f32 [[RESULT:v[0-9]+]], -[[TRUNC]], [[Y]], [[X]]
33 ; GCN: buffer_store_dword [[RESULT]]
35 define void @unsafe_frem_f32(float addrspace(1)* %out, float addrspace(1)* %in1,
36 float addrspace(1)* %in2) #1 {
37 %gep2 = getelementptr float addrspace(1)* %in2, i32 4
38 %r0 = load float addrspace(1)* %in1, align 4
39 %r1 = load float addrspace(1)* %gep2, align 4
40 %r2 = frem float %r0, %r1
41 store float %r2, float addrspace(1)* %out, align 4
46 ; FUNC-LABEL: {{^}}frem_f64:
47 ; GCN: buffer_load_dwordx2 [[Y:v\[[0-9]+:[0-9]+\]]], {{.*}}, 0
48 ; GCN: buffer_load_dwordx2 [[X:v\[[0-9]+:[0-9]+\]]], {{.*}}, 0
50 ; CI: v_rcp_f64_e32 [[INVY:v\[[0-9]+:[0-9]+\]]], [[Y]]
51 ; CI: v_mul_f64 [[DIV:v\[[0-9]+:[0-9]+\]]], [[X]], [[INVY]]
52 ; CI: v_trunc_f64_e32 [[TRUNC:v\[[0-9]+:[0-9]+\]]], [[DIV]]
53 ; CI: v_mul_f64 [[RESULTM:v\[[0-9]+:[0-9]+\]]], [[TRUNC]], [[Y]]
54 ; SI: v_mul_f64 [[RESULTM:v\[[0-9]+:[0-9]+\]]], {{v\[[0-9]+:[0-9]+\]}}, [[Y]]
55 ; GCN: v_add_f64 [[RESULT:v\[[0-9]+:[0-9]+\]]], [[X]], -[[RESULTM]]
56 ; GCN: buffer_store_dwordx2 [[RESULT]], {{.*}}, 0
58 define void @frem_f64(double addrspace(1)* %out, double addrspace(1)* %in1,
59 double addrspace(1)* %in2) #0 {
60 %r0 = load double addrspace(1)* %in1, align 8
61 %r1 = load double addrspace(1)* %in2, align 8
62 %r2 = frem double %r0, %r1
63 store double %r2, double addrspace(1)* %out, align 8
67 ; FUNC-LABEL: {{^}}unsafe_frem_f64:
74 define void @unsafe_frem_f64(double addrspace(1)* %out, double addrspace(1)* %in1,
75 double addrspace(1)* %in2) #1 {
76 %r0 = load double addrspace(1)* %in1, align 8
77 %r1 = load double addrspace(1)* %in2, align 8
78 %r2 = frem double %r0, %r1
79 store double %r2, double addrspace(1)* %out, align 8
83 define void @frem_v2f32(<2 x float> addrspace(1)* %out, <2 x float> addrspace(1)* %in1,
84 <2 x float> addrspace(1)* %in2) #0 {
85 %gep2 = getelementptr <2 x float> addrspace(1)* %in2, i32 4
86 %r0 = load <2 x float> addrspace(1)* %in1, align 8
87 %r1 = load <2 x float> addrspace(1)* %gep2, align 8
88 %r2 = frem <2 x float> %r0, %r1
89 store <2 x float> %r2, <2 x float> addrspace(1)* %out, align 8
93 define void @frem_v4f32(<4 x float> addrspace(1)* %out, <4 x float> addrspace(1)* %in1,
94 <4 x float> addrspace(1)* %in2) #0 {
95 %gep2 = getelementptr <4 x float> addrspace(1)* %in2, i32 4
96 %r0 = load <4 x float> addrspace(1)* %in1, align 16
97 %r1 = load <4 x float> addrspace(1)* %gep2, align 16
98 %r2 = frem <4 x float> %r0, %r1
99 store <4 x float> %r2, <4 x float> addrspace(1)* %out, align 16
103 define void @frem_v2f64(<2 x double> addrspace(1)* %out, <2 x double> addrspace(1)* %in1,
104 <2 x double> addrspace(1)* %in2) #0 {
105 %gep2 = getelementptr <2 x double> addrspace(1)* %in2, i32 4
106 %r0 = load <2 x double> addrspace(1)* %in1, align 16
107 %r1 = load <2 x double> addrspace(1)* %gep2, align 16
108 %r2 = frem <2 x double> %r0, %r1
109 store <2 x double> %r2, <2 x double> addrspace(1)* %out, align 16
113 attributes #0 = { nounwind "unsafe-fp-math"="false" }
114 attributes #1 = { nounwind "unsafe-fp-math"="true" }