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
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn-amd-amdpal -mattr=-real-true16 -mcpu=gfx1100 -o - %s | FileCheck -check-prefixes=GCN,GFX11,GFX11-FAKE16 %s
; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn-amd-amdpal -mattr=+real-true16 -mcpu=gfx1100 -o - %s | FileCheck -check-prefixes=GCN,GFX11,GFX11-TRUE16 %s
; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn-amd-amdpal -mattr=-real-true16 -mcpu=gfx1200 -o - %s | FileCheck -check-prefixes=GCN,GFX12,GFX12-FAKE16 %s
; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn-amd-amdpal -mattr=+real-true16 -mcpu=gfx1200 -o - %s | FileCheck -check-prefixes=GCN,GFX12,GFX12-TRUE16 %s
define amdgpu_ps half @fadd_s16_uniform(half inreg %a, half inreg %b) {
; GFX11-FAKE16-LABEL: fadd_s16_uniform:
; GFX11-FAKE16: ; %bb.0:
; GFX11-FAKE16-NEXT: v_add_f16_e64 v0, s0, s1
; GFX11-FAKE16-NEXT: ; return to shader part epilog
;
; GFX11-TRUE16-LABEL: fadd_s16_uniform:
; GFX11-TRUE16: ; %bb.0:
; GFX11-TRUE16-NEXT: v_add_f16_e64 v0.l, s0, s1
; GFX11-TRUE16-NEXT: ; return to shader part epilog
;
; GFX12-LABEL: fadd_s16_uniform:
; GFX12: ; %bb.0:
; GFX12-NEXT: s_add_f16 s0, s0, s1
; GFX12-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; GFX12-NEXT: v_mov_b32_e32 v0, s0
; GFX12-NEXT: ; return to shader part epilog
%fadd = fadd half %a, %b
ret half %fadd
}
define amdgpu_ps half @fadd_s16_div(half %a, half %b) {
; GFX11-FAKE16-LABEL: fadd_s16_div:
; GFX11-FAKE16: ; %bb.0:
; GFX11-FAKE16-NEXT: v_add_f16_e32 v0, v0, v1
; GFX11-FAKE16-NEXT: ; return to shader part epilog
;
; GFX11-TRUE16-LABEL: fadd_s16_div:
; GFX11-TRUE16: ; %bb.0:
; GFX11-TRUE16-NEXT: v_add_f16_e32 v0.l, v0.l, v1.l
; GFX11-TRUE16-NEXT: ; return to shader part epilog
;
; GFX12-FAKE16-LABEL: fadd_s16_div:
; GFX12-FAKE16: ; %bb.0:
; GFX12-FAKE16-NEXT: v_add_f16_e32 v0, v0, v1
; GFX12-FAKE16-NEXT: ; return to shader part epilog
;
; GFX12-TRUE16-LABEL: fadd_s16_div:
; GFX12-TRUE16: ; %bb.0:
; GFX12-TRUE16-NEXT: v_add_f16_e32 v0.l, v0.l, v1.l
; GFX12-TRUE16-NEXT: ; return to shader part epilog
%fadd = fadd half %a, %b
ret half %fadd
}
define amdgpu_ps float @fadd_s32_uniform(float inreg %a, float inreg %b) {
; GFX11-LABEL: fadd_s32_uniform:
; GFX11: ; %bb.0:
; GFX11-NEXT: v_add_f32_e64 v0, s0, s1
; GFX11-NEXT: ; return to shader part epilog
;
; GFX12-LABEL: fadd_s32_uniform:
; GFX12: ; %bb.0:
; GFX12-NEXT: s_add_f32 s0, s0, s1
; GFX12-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; GFX12-NEXT: v_mov_b32_e32 v0, s0
; GFX12-NEXT: ; return to shader part epilog
%fadd = fadd float %a, %b
ret float %fadd
}
define amdgpu_ps float @fadd_s32_div(float %a, float %b) {
; GCN-LABEL: fadd_s32_div:
; GCN: ; %bb.0:
; GCN-NEXT: v_add_f32_e32 v0, v0, v1
; GCN-NEXT: ; return to shader part epilog
%fadd = fadd float %a, %b
ret float %fadd
}
define amdgpu_ps void @fadd_s64_uniform(double inreg %a, double inreg %b, ptr addrspace(1) %ptr) {
; GFX11-LABEL: fadd_s64_uniform:
; GFX11: ; %bb.0:
; GFX11-NEXT: v_add_f64 v[2:3], s[0:1], s[2:3]
; GFX11-NEXT: global_store_b64 v[0:1], v[2:3], off
; GFX11-NEXT: s_endpgm
;
; GFX12-LABEL: fadd_s64_uniform:
; GFX12: ; %bb.0:
; GFX12-NEXT: v_add_f64_e64 v[2:3], s[0:1], s[2:3]
; GFX12-NEXT: global_store_b64 v[0:1], v[2:3], off
; GFX12-NEXT: s_endpgm
%fadd = fadd double %a, %b
store double %fadd, ptr addrspace(1) %ptr
ret void
}
define amdgpu_ps void @fadd_s64_div(double %a, double %b, ptr addrspace(1) %ptr) {
; GFX11-LABEL: fadd_s64_div:
; GFX11: ; %bb.0:
; GFX11-NEXT: v_add_f64 v[0:1], v[0:1], v[2:3]
; GFX11-NEXT: global_store_b64 v[4:5], v[0:1], off
; GFX11-NEXT: s_endpgm
;
; GFX12-LABEL: fadd_s64_div:
; GFX12: ; %bb.0:
; GFX12-NEXT: v_add_f64_e32 v[0:1], v[0:1], v[2:3]
; GFX12-NEXT: global_store_b64 v[4:5], v[0:1], off
; GFX12-NEXT: s_endpgm
%fadd = fadd double %a, %b
store double %fadd, ptr addrspace(1) %ptr
ret void
}
define amdgpu_ps <2 x half> @fadd_v2s16_uniform(<2 x half> inreg %a, <2 x half> inreg %b) {
; GFX11-LABEL: fadd_v2s16_uniform:
; GFX11: ; %bb.0:
; GFX11-NEXT: v_pk_add_f16 v0, s0, s1
; GFX11-NEXT: ; return to shader part epilog
;
; GFX12-LABEL: fadd_v2s16_uniform:
; GFX12: ; %bb.0:
; GFX12-NEXT: s_lshr_b32 s2, s0, 16
; GFX12-NEXT: s_lshr_b32 s3, s1, 16
; GFX12-NEXT: s_add_f16 s0, s0, s1
; GFX12-NEXT: s_add_f16 s1, s2, s3
; GFX12-NEXT: s_delay_alu instid0(SALU_CYCLE_3) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX12-NEXT: s_pack_ll_b32_b16 s0, s0, s1
; GFX12-NEXT: v_mov_b32_e32 v0, s0
; GFX12-NEXT: ; return to shader part epilog
%fadd = fadd <2 x half> %a, %b
ret <2 x half> %fadd
}
define amdgpu_ps <2 x half> @fadd_v2s16_div(<2 x half> %a, <2 x half> %b) {
; GCN-LABEL: fadd_v2s16_div:
; GCN: ; %bb.0:
; GCN-NEXT: v_pk_add_f16 v0, v0, v1
; GCN-NEXT: ; return to shader part epilog
%fadd = fadd <2 x half> %a, %b
ret <2 x half> %fadd
}
define amdgpu_ps <2 x float> @fadd_v2s32_uniform(<2 x float> inreg %a, <2 x float> inreg %b) {
; GFX11-LABEL: fadd_v2s32_uniform:
; GFX11: ; %bb.0:
; GFX11-NEXT: v_add_f32_e64 v0, s0, s2
; GFX11-NEXT: v_add_f32_e64 v1, s1, s3
; GFX11-NEXT: ; return to shader part epilog
;
; GFX12-LABEL: fadd_v2s32_uniform:
; GFX12: ; %bb.0:
; GFX12-NEXT: s_add_f32 s0, s0, s2
; GFX12-NEXT: s_add_f32 s1, s1, s3
; GFX12-NEXT: s_delay_alu instid0(SALU_CYCLE_3)
; GFX12-NEXT: v_dual_mov_b32 v0, s0 :: v_dual_mov_b32 v1, s1
; GFX12-NEXT: ; return to shader part epilog
%fadd = fadd <2 x float> %a, %b
ret <2 x float> %fadd
}
define amdgpu_ps <2 x float> @fadd_v2s32_div(<2 x float> %a, <2 x float> %b) {
; GCN-LABEL: fadd_v2s32_div:
; GCN: ; %bb.0:
; GCN-NEXT: v_dual_add_f32 v0, v0, v2 :: v_dual_add_f32 v1, v1, v3
; GCN-NEXT: ; return to shader part epilog
%fadd = fadd <2 x float> %a, %b
ret <2 x float> %fadd
}
|
