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
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
; RUN: llc -march=amdgcn -mcpu=gfx803 -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=GFX8 %s
; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=GFX9 %s
; RUN: llc -march=amdgcn -mcpu=gfx1010 -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=GFX10 %s
; RUN: llc -march=amdgcn -mcpu=gfx1100 -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=GFX11 %s
; FIXME: GFX9 should be producing v_mad_u16 instead of v_mad_legacy_u16.
define amdgpu_kernel void @mad_u16(
; GFX8-LABEL: mad_u16:
; GFX8: ; %bb.0: ; %entry
; GFX8-NEXT: s_load_dwordx8 s[0:7], s[0:1], 0x24
; GFX8-NEXT: v_lshlrev_b32_e32 v4, 1, v0
; GFX8-NEXT: s_waitcnt lgkmcnt(0)
; GFX8-NEXT: v_mov_b32_e32 v1, s3
; GFX8-NEXT: v_add_u32_e32 v0, vcc, s2, v4
; GFX8-NEXT: v_addc_u32_e32 v1, vcc, 0, v1, vcc
; GFX8-NEXT: v_mov_b32_e32 v3, s5
; GFX8-NEXT: v_add_u32_e32 v2, vcc, s4, v4
; GFX8-NEXT: v_addc_u32_e32 v3, vcc, 0, v3, vcc
; GFX8-NEXT: v_mov_b32_e32 v5, s7
; GFX8-NEXT: v_add_u32_e32 v4, vcc, s6, v4
; GFX8-NEXT: v_addc_u32_e32 v5, vcc, 0, v5, vcc
; GFX8-NEXT: flat_load_ushort v6, v[0:1] glc
; GFX8-NEXT: s_waitcnt vmcnt(0)
; GFX8-NEXT: flat_load_ushort v2, v[2:3] glc
; GFX8-NEXT: s_waitcnt vmcnt(0)
; GFX8-NEXT: flat_load_ushort v3, v[4:5] glc
; GFX8-NEXT: s_waitcnt vmcnt(0)
; GFX8-NEXT: v_mov_b32_e32 v0, s0
; GFX8-NEXT: v_mov_b32_e32 v1, s1
; GFX8-NEXT: v_mad_u16 v2, v6, v2, v3
; GFX8-NEXT: flat_store_short v[0:1], v2
; GFX8-NEXT: s_endpgm
;
; GFX9-LABEL: mad_u16:
; GFX9: ; %bb.0: ; %entry
; GFX9-NEXT: s_load_dwordx8 s[0:7], s[0:1], 0x24
; GFX9-NEXT: v_lshlrev_b32_e32 v0, 1, v0
; GFX9-NEXT: s_waitcnt lgkmcnt(0)
; GFX9-NEXT: global_load_ushort v1, v0, s[2:3] glc
; GFX9-NEXT: s_waitcnt vmcnt(0)
; GFX9-NEXT: global_load_ushort v2, v0, s[4:5] glc
; GFX9-NEXT: s_waitcnt vmcnt(0)
; GFX9-NEXT: global_load_ushort v3, v0, s[6:7] glc
; GFX9-NEXT: s_waitcnt vmcnt(0)
; GFX9-NEXT: v_mov_b32_e32 v0, 0
; GFX9-NEXT: v_mad_legacy_u16 v1, v1, v2, v3
; GFX9-NEXT: global_store_short v0, v1, s[0:1]
; GFX9-NEXT: s_endpgm
;
; GFX10-LABEL: mad_u16:
; GFX10: ; %bb.0: ; %entry
; GFX10-NEXT: s_load_dwordx8 s[0:7], s[0:1], 0x24
; GFX10-NEXT: v_lshlrev_b32_e32 v0, 1, v0
; GFX10-NEXT: s_waitcnt lgkmcnt(0)
; GFX10-NEXT: global_load_ushort v1, v0, s[2:3] glc dlc
; GFX10-NEXT: s_waitcnt vmcnt(0)
; GFX10-NEXT: global_load_ushort v2, v0, s[4:5] glc dlc
; GFX10-NEXT: s_waitcnt vmcnt(0)
; GFX10-NEXT: global_load_ushort v3, v0, s[6:7] glc dlc
; GFX10-NEXT: s_waitcnt vmcnt(0)
; GFX10-NEXT: v_mov_b32_e32 v0, 0
; GFX10-NEXT: v_mad_u16 v1, v1, v2, v3
; GFX10-NEXT: global_store_short v0, v1, s[0:1]
; GFX10-NEXT: s_endpgm
;
; GFX11-LABEL: mad_u16:
; GFX11: ; %bb.0: ; %entry
; GFX11-NEXT: s_load_b256 s[0:7], s[0:1], 0x24
; GFX11-NEXT: v_dual_mov_b32 v3, 0 :: v_dual_lshlrev_b32 v0, 1, v0
; GFX11-NEXT: s_waitcnt lgkmcnt(0)
; GFX11-NEXT: global_load_u16 v1, v0, s[2:3] glc dlc
; GFX11-NEXT: s_waitcnt vmcnt(0)
; GFX11-NEXT: global_load_u16 v2, v0, s[4:5] glc dlc
; GFX11-NEXT: s_waitcnt vmcnt(0)
; GFX11-NEXT: global_load_u16 v0, v0, s[6:7] glc dlc
; GFX11-NEXT: s_waitcnt vmcnt(0)
; GFX11-NEXT: v_mad_u16 v0, v1, v2, v0
; GFX11-NEXT: global_store_b16 v3, v0, s[0:1]
; GFX11-NEXT: s_nop 0
; GFX11-NEXT: s_sendmsg sendmsg(MSG_DEALLOC_VGPRS)
; GFX11-NEXT: s_endpgm
ptr addrspace(1) %r,
ptr addrspace(1) %a,
ptr addrspace(1) %b,
ptr addrspace(1) %c) {
entry:
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%a.gep = getelementptr inbounds i16, ptr addrspace(1) %a, i32 %tid
%b.gep = getelementptr inbounds i16, ptr addrspace(1) %b, i32 %tid
%c.gep = getelementptr inbounds i16, ptr addrspace(1) %c, i32 %tid
%a.val = load volatile i16, ptr addrspace(1) %a.gep
%b.val = load volatile i16, ptr addrspace(1) %b.gep
%c.val = load volatile i16, ptr addrspace(1) %c.gep
%m.val = mul i16 %a.val, %b.val
%r.val = add i16 %m.val, %c.val
store i16 %r.val, ptr addrspace(1) %r
ret void
}
define i16 @v_mad_u16(i16 %arg0, i16 %arg1, i16 %arg2) {
; GFX8-LABEL: v_mad_u16:
; GFX8: ; %bb.0:
; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX8-NEXT: v_mad_u16 v0, v0, v1, v2
; GFX8-NEXT: s_setpc_b64 s[30:31]
;
; GFX9-LABEL: v_mad_u16:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9-NEXT: v_mad_legacy_u16 v0, v0, v1, v2
; GFX9-NEXT: s_setpc_b64 s[30:31]
;
; GFX10-LABEL: v_mad_u16:
; GFX10: ; %bb.0:
; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10-NEXT: v_mad_u16 v0, v0, v1, v2
; GFX10-NEXT: s_setpc_b64 s[30:31]
;
; GFX11-LABEL: v_mad_u16:
; GFX11: ; %bb.0:
; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX11-NEXT: v_mad_u16 v0, v0, v1, v2
; GFX11-NEXT: s_setpc_b64 s[30:31]
%mul = mul i16 %arg0, %arg1
%add = add i16 %mul, %arg2
ret i16 %add
}
define i32 @v_mad_u16_zext(i16 %arg0, i16 %arg1, i16 %arg2) {
; GFX8-LABEL: v_mad_u16_zext:
; GFX8: ; %bb.0:
; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX8-NEXT: v_mad_u16 v0, v0, v1, v2
; GFX8-NEXT: s_setpc_b64 s[30:31]
;
; GFX9-LABEL: v_mad_u16_zext:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9-NEXT: v_mad_legacy_u16 v0, v0, v1, v2
; GFX9-NEXT: s_setpc_b64 s[30:31]
;
; GFX10-LABEL: v_mad_u16_zext:
; GFX10: ; %bb.0:
; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10-NEXT: v_mad_u16 v0, v0, v1, v2
; GFX10-NEXT: v_and_b32_e32 v0, 0xffff, v0
; GFX10-NEXT: s_setpc_b64 s[30:31]
;
; GFX11-LABEL: v_mad_u16_zext:
; GFX11: ; %bb.0:
; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX11-NEXT: v_mad_u16 v0, v0, v1, v2
; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX11-NEXT: v_and_b32_e32 v0, 0xffff, v0
; GFX11-NEXT: s_setpc_b64 s[30:31]
%mul = mul i16 %arg0, %arg1
%add = add i16 %mul, %arg2
%zext = zext i16 %add to i32
ret i32 %zext
}
define i64 @v_mad_u16_zext64(i16 %arg0, i16 %arg1, i16 %arg2) {
; GFX8-LABEL: v_mad_u16_zext64:
; GFX8: ; %bb.0:
; GFX8-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX8-NEXT: v_mad_u16 v0, v0, v1, v2
; GFX8-NEXT: v_mov_b32_e32 v1, 0
; GFX8-NEXT: s_setpc_b64 s[30:31]
;
; GFX9-LABEL: v_mad_u16_zext64:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9-NEXT: v_mad_legacy_u16 v0, v0, v1, v2
; GFX9-NEXT: v_mov_b32_e32 v1, 0
; GFX9-NEXT: s_setpc_b64 s[30:31]
;
; GFX10-LABEL: v_mad_u16_zext64:
; GFX10: ; %bb.0:
; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10-NEXT: v_mad_u16 v0, v0, v1, v2
; GFX10-NEXT: v_mov_b32_e32 v1, 0
; GFX10-NEXT: v_and_b32_e32 v0, 0xffff, v0
; GFX10-NEXT: s_setpc_b64 s[30:31]
;
; GFX11-LABEL: v_mad_u16_zext64:
; GFX11: ; %bb.0:
; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX11-NEXT: v_mad_u16 v0, v0, v1, v2
; GFX11-NEXT: s_delay_alu instid0(VALU_DEP_1)
; GFX11-NEXT: v_dual_mov_b32 v1, 0 :: v_dual_and_b32 v0, 0xffff, v0
; GFX11-NEXT: s_setpc_b64 s[30:31]
%mul = mul i16 %arg0, %arg1
%add = add i16 %mul, %arg2
%zext = zext i16 %add to i64
ret i64 %zext
}
define amdgpu_ps i16 @s_mad_u16(i16 inreg %arg0, i16 inreg %arg1, i16 inreg %arg2) {
; GFX8-LABEL: s_mad_u16:
; GFX8: ; %bb.0:
; GFX8-NEXT: s_mul_i32 s0, s0, s1
; GFX8-NEXT: s_add_i32 s0, s0, s2
; GFX8-NEXT: ; return to shader part epilog
;
; GFX9-LABEL: s_mad_u16:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_mul_i32 s0, s0, s1
; GFX9-NEXT: s_add_i32 s0, s0, s2
; GFX9-NEXT: ; return to shader part epilog
;
; GFX10-LABEL: s_mad_u16:
; GFX10: ; %bb.0:
; GFX10-NEXT: s_mul_i32 s0, s0, s1
; GFX10-NEXT: s_add_i32 s0, s0, s2
; GFX10-NEXT: ; return to shader part epilog
;
; GFX11-LABEL: s_mad_u16:
; GFX11: ; %bb.0:
; GFX11-NEXT: s_mul_i32 s0, s0, s1
; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-NEXT: s_add_i32 s0, s0, s2
; GFX11-NEXT: ; return to shader part epilog
%mul = mul i16 %arg0, %arg1
%add = add i16 %mul, %arg2
ret i16 %add
}
define amdgpu_ps i32 @s_mad_u16_zext(i16 inreg %arg0, i16 inreg %arg1, i16 inreg %arg2) {
; GFX8-LABEL: s_mad_u16_zext:
; GFX8: ; %bb.0:
; GFX8-NEXT: s_mul_i32 s0, s0, s1
; GFX8-NEXT: s_add_i32 s0, s0, s2
; GFX8-NEXT: s_and_b32 s0, s0, 0xffff
; GFX8-NEXT: ; return to shader part epilog
;
; GFX9-LABEL: s_mad_u16_zext:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_mul_i32 s0, s0, s1
; GFX9-NEXT: s_add_i32 s0, s0, s2
; GFX9-NEXT: s_and_b32 s0, s0, 0xffff
; GFX9-NEXT: ; return to shader part epilog
;
; GFX10-LABEL: s_mad_u16_zext:
; GFX10: ; %bb.0:
; GFX10-NEXT: s_mul_i32 s0, s0, s1
; GFX10-NEXT: s_add_i32 s0, s0, s2
; GFX10-NEXT: s_and_b32 s0, s0, 0xffff
; GFX10-NEXT: ; return to shader part epilog
;
; GFX11-LABEL: s_mad_u16_zext:
; GFX11: ; %bb.0:
; GFX11-NEXT: s_mul_i32 s0, s0, s1
; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX11-NEXT: s_add_i32 s0, s0, s2
; GFX11-NEXT: s_and_b32 s0, s0, 0xffff
; GFX11-NEXT: ; return to shader part epilog
%mul = mul i16 %arg0, %arg1
%add = add i16 %mul, %arg2
%zext = zext i16 %add to i32
ret i32 %zext
}
define amdgpu_ps i64 @s_mad_u16_zext64(i16 inreg %arg0, i16 inreg %arg1, i16 inreg %arg2) {
; GFX8-LABEL: s_mad_u16_zext64:
; GFX8: ; %bb.0:
; GFX8-NEXT: s_mul_i32 s0, s0, s1
; GFX8-NEXT: s_add_i32 s0, s0, s2
; GFX8-NEXT: s_and_b32 s0, s0, 0xffff
; GFX8-NEXT: s_mov_b32 s1, 0
; GFX8-NEXT: ; return to shader part epilog
;
; GFX9-LABEL: s_mad_u16_zext64:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_mul_i32 s0, s0, s1
; GFX9-NEXT: s_add_i32 s0, s0, s2
; GFX9-NEXT: s_and_b32 s0, s0, 0xffff
; GFX9-NEXT: s_mov_b32 s1, 0
; GFX9-NEXT: ; return to shader part epilog
;
; GFX10-LABEL: s_mad_u16_zext64:
; GFX10: ; %bb.0:
; GFX10-NEXT: s_mul_i32 s0, s0, s1
; GFX10-NEXT: s_mov_b32 s1, 0
; GFX10-NEXT: s_add_i32 s0, s0, s2
; GFX10-NEXT: s_and_b32 s0, s0, 0xffff
; GFX10-NEXT: ; return to shader part epilog
;
; GFX11-LABEL: s_mad_u16_zext64:
; GFX11: ; %bb.0:
; GFX11-NEXT: s_mul_i32 s0, s0, s1
; GFX11-NEXT: s_mov_b32 s1, 0
; GFX11-NEXT: s_add_i32 s0, s0, s2
; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-NEXT: s_and_b32 s0, s0, 0xffff
; GFX11-NEXT: ; return to shader part epilog
%mul = mul i16 %arg0, %arg1
%add = add i16 %mul, %arg2
%zext = zext i16 %add to i64
ret i64 %zext
}
define amdgpu_ps i32 @s_mad_u16_sext(i16 inreg %arg0, i16 inreg %arg1, i16 inreg %arg2) {
; GFX8-LABEL: s_mad_u16_sext:
; GFX8: ; %bb.0:
; GFX8-NEXT: s_mul_i32 s0, s0, s1
; GFX8-NEXT: s_add_i32 s0, s0, s2
; GFX8-NEXT: s_sext_i32_i16 s0, s0
; GFX8-NEXT: ; return to shader part epilog
;
; GFX9-LABEL: s_mad_u16_sext:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_mul_i32 s0, s0, s1
; GFX9-NEXT: s_add_i32 s0, s0, s2
; GFX9-NEXT: s_sext_i32_i16 s0, s0
; GFX9-NEXT: ; return to shader part epilog
;
; GFX10-LABEL: s_mad_u16_sext:
; GFX10: ; %bb.0:
; GFX10-NEXT: s_mul_i32 s0, s0, s1
; GFX10-NEXT: s_add_i32 s0, s0, s2
; GFX10-NEXT: s_sext_i32_i16 s0, s0
; GFX10-NEXT: ; return to shader part epilog
;
; GFX11-LABEL: s_mad_u16_sext:
; GFX11: ; %bb.0:
; GFX11-NEXT: s_mul_i32 s0, s0, s1
; GFX11-NEXT: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX11-NEXT: s_add_i32 s0, s0, s2
; GFX11-NEXT: s_sext_i32_i16 s0, s0
; GFX11-NEXT: ; return to shader part epilog
%mul = mul i16 %arg0, %arg1
%add = add i16 %mul, %arg2
%sext = sext i16 %add to i32
ret i32 %sext
}
declare i32 @llvm.amdgcn.workitem.id.x()
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
; GCN: {{.*}}
|
