1 ; RUN: llc -march=amdgcn -mcpu=SI -verify-machineinstrs -enable-unsafe-fp-math < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
2 ; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs -enable-unsafe-fp-math < %s | FileCheck -check-prefix=CI -check-prefix=FUNC %s
3 ; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs -enable-unsafe-fp-math < %s | FileCheck -check-prefix=CI -check-prefix=FUNC %s
5 declare double @llvm.fabs.f64(double %Val)
6 declare double @llvm.floor.f64(double) nounwind readnone
7 declare <2 x double> @llvm.floor.v2f64(<2 x double>) nounwind readnone
8 declare <3 x double> @llvm.floor.v3f64(<3 x double>) nounwind readnone
9 declare <4 x double> @llvm.floor.v4f64(<4 x double>) nounwind readnone
10 declare <8 x double> @llvm.floor.v8f64(<8 x double>) nounwind readnone
11 declare <16 x double> @llvm.floor.v16f64(<16 x double>) nounwind readnone
13 ; FUNC-LABEL: {{^}}ffloor_f64:
17 ; SI: v_cmp_class_f64_e64
18 ; SI: v_cndmask_b32_e64
19 ; SI: v_cndmask_b32_e64
22 define void @ffloor_f64(double addrspace(1)* %out, double %x) {
23 %y = call double @llvm.floor.f64(double %x) nounwind readnone
24 store double %y, double addrspace(1)* %out
28 ; FUNC-LABEL: {{^}}ffloor_f64_neg:
30 ; SI: v_fract_f64_e64 {{v[[0-9]+:[0-9]+]}}, -[[INPUT:s[[0-9]+:[0-9]+]]]
32 ; SI: v_cmp_class_f64_e64
33 ; SI: v_cndmask_b32_e64
34 ; SI: v_cndmask_b32_e64
35 ; SI: v_add_f64 {{v[[0-9]+:[0-9]+]}}, -[[INPUT]]
37 define void @ffloor_f64_neg(double addrspace(1)* %out, double %x) {
38 %neg = fsub double 0.0, %x
39 %y = call double @llvm.floor.f64(double %neg) nounwind readnone
40 store double %y, double addrspace(1)* %out
44 ; FUNC-LABEL: {{^}}ffloor_f64_neg_abs:
46 ; SI: v_fract_f64_e64 {{v[[0-9]+:[0-9]+]}}, -|[[INPUT:s[[0-9]+:[0-9]+]]]|
48 ; SI: v_cmp_class_f64_e64
49 ; SI: v_cndmask_b32_e64
50 ; SI: v_cndmask_b32_e64
51 ; SI: v_add_f64 {{v[[0-9]+:[0-9]+]}}, -|[[INPUT]]|
53 define void @ffloor_f64_neg_abs(double addrspace(1)* %out, double %x) {
54 %abs = call double @llvm.fabs.f64(double %x)
55 %neg = fsub double 0.0, %abs
56 %y = call double @llvm.floor.f64(double %neg) nounwind readnone
57 store double %y, double addrspace(1)* %out
61 ; FUNC-LABEL: {{^}}ffloor_v2f64:
64 define void @ffloor_v2f64(<2 x double> addrspace(1)* %out, <2 x double> %x) {
65 %y = call <2 x double> @llvm.floor.v2f64(<2 x double> %x) nounwind readnone
66 store <2 x double> %y, <2 x double> addrspace(1)* %out
70 ; FIXME-FUNC-LABEL: {{^}}ffloor_v3f64:
71 ; FIXME-CI: v_floor_f64_e32
72 ; FIXME-CI: v_floor_f64_e32
73 ; FIXME-CI: v_floor_f64_e32
74 ; define void @ffloor_v3f64(<3 x double> addrspace(1)* %out, <3 x double> %x) {
75 ; %y = call <3 x double> @llvm.floor.v3f64(<3 x double> %x) nounwind readnone
76 ; store <3 x double> %y, <3 x double> addrspace(1)* %out
80 ; FUNC-LABEL: {{^}}ffloor_v4f64:
85 define void @ffloor_v4f64(<4 x double> addrspace(1)* %out, <4 x double> %x) {
86 %y = call <4 x double> @llvm.floor.v4f64(<4 x double> %x) nounwind readnone
87 store <4 x double> %y, <4 x double> addrspace(1)* %out
91 ; FUNC-LABEL: {{^}}ffloor_v8f64:
100 define void @ffloor_v8f64(<8 x double> addrspace(1)* %out, <8 x double> %x) {
101 %y = call <8 x double> @llvm.floor.v8f64(<8 x double> %x) nounwind readnone
102 store <8 x double> %y, <8 x double> addrspace(1)* %out
106 ; FUNC-LABEL: {{^}}ffloor_v16f64:
107 ; CI: v_floor_f64_e32
108 ; CI: v_floor_f64_e32
109 ; CI: v_floor_f64_e32
110 ; CI: v_floor_f64_e32
111 ; CI: v_floor_f64_e32
112 ; CI: v_floor_f64_e32
113 ; CI: v_floor_f64_e32
114 ; CI: v_floor_f64_e32
115 ; CI: v_floor_f64_e32
116 ; CI: v_floor_f64_e32
117 ; CI: v_floor_f64_e32
118 ; CI: v_floor_f64_e32
119 ; CI: v_floor_f64_e32
120 ; CI: v_floor_f64_e32
121 ; CI: v_floor_f64_e32
122 ; CI: v_floor_f64_e32
123 define void @ffloor_v16f64(<16 x double> addrspace(1)* %out, <16 x double> %x) {
124 %y = call <16 x double> @llvm.floor.v16f64(<16 x double> %x) nounwind readnone
125 store <16 x double> %y, <16 x double> addrspace(1)* %out