1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=amdgcn -mcpu=gfx1150 -verify-machineinstrs < %s | FileCheck %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1150 -global-isel -verify-machineinstrs < %s | FileCheck %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -verify-machineinstrs < %s | FileCheck %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -global-isel -verify-machineinstrs < %s | FileCheck %s
define amdgpu_vs float @sitofp_i32_to_f32(i32 inreg %val) {
; CHECK-LABEL: sitofp_i32_to_f32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_cvt_f32_i32 s0, s0
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; CHECK-NEXT: v_mov_b32_e32 v0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = sitofp i32 %val to float
ret float %res
}
define amdgpu_vs float @uitofp_u32_to_f32(i32 inreg %val) {
; CHECK-LABEL: uitofp_u32_to_f32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_cvt_f32_u32 s0, s0
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; CHECK-NEXT: v_mov_b32_e32 v0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = uitofp i32 %val to float
ret float %res
}
define amdgpu_vs i32 @fptosi_f32_to_i32(float inreg %val) {
; CHECK-LABEL: fptosi_f32_to_i32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_cvt_i32_f32 s0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = fptosi float %val to i32
ret i32 %res
}
define amdgpu_vs i32 @fptoui_f32_to_u32(float inreg %val) {
; CHECK-LABEL: fptoui_f32_to_u32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_cvt_u32_f32 s0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = fptoui float %val to i32
ret i32 %res
}
define amdgpu_vs float @fpext_f16_to_f32(half inreg %val) {
; CHECK-LABEL: fpext_f16_to_f32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_cvt_f32_f16 s0, s0
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; CHECK-NEXT: v_mov_b32_e32 v0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = fpext half %val to float
ret float %res
}
define amdgpu_vs float @fpext_hif16_to_32(<2 x half> inreg %val) {
; CHECK-LABEL: fpext_hif16_to_32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_cvt_hi_f32_f16 s0, s0
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; CHECK-NEXT: v_mov_b32_e32 v0, s0
; CHECK-NEXT: ; return to shader part epilog
%hielt = extractelement <2 x half> %val, i32 1
%res = fpext half %hielt to float
ret float %res
}
define amdgpu_vs half @fptrunc_f32_to_f16(float inreg %val) {
; CHECK-LABEL: fptrunc_f32_to_f16:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_cvt_f16_f32 s0, s0
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; CHECK-NEXT: v_mov_b32_e32 v0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = fptrunc float %val to half
ret half %res
}
define amdgpu_vs float @fceil_f32(float inreg %val) {
; CHECK-LABEL: fceil_f32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_ceil_f32 s0, s0
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; CHECK-NEXT: v_mov_b32_e32 v0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = call float @llvm.ceil.f32(float %val)
ret float %res
}
define amdgpu_vs float @ffloor_f32(float inreg %val) {
; CHECK-LABEL: ffloor_f32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_floor_f32 s0, s0
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; CHECK-NEXT: v_mov_b32_e32 v0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = call float @llvm.floor.f32(float %val)
ret float %res
}
define amdgpu_vs float @ftrunc_f32(float inreg %val) {
; CHECK-LABEL: ftrunc_f32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_trunc_f32 s0, s0
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; CHECK-NEXT: v_mov_b32_e32 v0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = call float @llvm.trunc.f32(float %val)
ret float %res
}
define amdgpu_vs float @frint_f32(float inreg %val) {
; CHECK-LABEL: frint_f32:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_rndne_f32 s0, s0
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; CHECK-NEXT: v_mov_b32_e32 v0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = call float @llvm.rint.f32(float %val)
ret float %res
}
define amdgpu_vs half @fceil_f16(half inreg %val) {
; CHECK-LABEL: fceil_f16:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_ceil_f16 s0, s0
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; CHECK-NEXT: v_mov_b32_e32 v0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = call half @llvm.ceil.f16(half %val)
ret half %res
}
define amdgpu_vs half @ffloor_f16(half inreg %val) {
; CHECK-LABEL: ffloor_f16:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_floor_f16 s0, s0
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; CHECK-NEXT: v_mov_b32_e32 v0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = call half @llvm.floor.f16(half %val)
ret half %res
}
define amdgpu_vs half @ftrunc_f16(half inreg %val) {
; CHECK-LABEL: ftrunc_f16:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_trunc_f16 s0, s0
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; CHECK-NEXT: v_mov_b32_e32 v0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = call half @llvm.trunc.f16(half %val)
ret half %res
}
define amdgpu_vs half @frint_f16(half inreg %val) {
; CHECK-LABEL: frint_f16:
; CHECK: ; %bb.0:
; CHECK-NEXT: s_rndne_f16 s0, s0
; CHECK-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; CHECK-NEXT: v_mov_b32_e32 v0, s0
; CHECK-NEXT: ; return to shader part epilog
%res = call half @llvm.rint.f16(half %val)
ret half %res
}
declare float @llvm.ceil.f32(float)
declare float @llvm.floor.f32(float)
declare float @llvm.trunc.f32(float)
declare float @llvm.rint.f32(float)
declare half @llvm.ceil.f16(half)
declare half @llvm.floor.f16(half)
declare half @llvm.trunc.f16(half)
declare half @llvm.rint.f16(half)
|
