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
|
; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
; RUN: llc -march=amdgcn -mcpu=gfx90a -verify-machineinstrs -stop-after=amdgpu-isel < %s | FileCheck -check-prefix=GFX90A_GFX940 %s
; RUN: llc -march=amdgcn -mcpu=gfx940 -verify-machineinstrs -stop-after=amdgpu-isel < %s | FileCheck -check-prefix=GFX90A_GFX940 %s
; RUN: llc -march=amdgcn -mcpu=gfx1100 -verify-machineinstrs -stop-after=amdgpu-isel < %s | FileCheck -check-prefix=GFX11 %s
define amdgpu_ps float @buffer_atomic_fadd_f32_offset_rtn(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offset_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFSET_RTN [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0
; GFX11-LABEL: name: buffer_atomic_fadd_f32_offset_rtn
; GFX11: bb.0 (%ir-block.0):
; GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4
; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX11-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3
; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFSET_RTN [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7)
; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG $vgpr0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
ret float %ret
}
define amdgpu_ps float @buffer_atomic_fadd_f32_offen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offen_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0
; GFX11-LABEL: name: buffer_atomic_fadd_f32_offen_rtn
; GFX11: bb.0 (%ir-block.0):
; GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7)
; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG $vgpr0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %ret
}
define amdgpu_ps float @buffer_atomic_fadd_f32_idxen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_idxen_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_IDXEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0
; GFX11-LABEL: name: buffer_atomic_fadd_f32_idxen_rtn
; GFX11: bb.0 (%ir-block.0):
; GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_IDXEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7)
; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG $vgpr0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret float %ret
}
define amdgpu_ps float @buffer_atomic_fadd_f32_bothen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_bothen_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0
; GFX11-LABEL: name: buffer_atomic_fadd_f32_bothen_rtn
; GFX11: bb.0 (%ir-block.0):
; GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4
; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX11-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX11-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX11-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3
; GFX11-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7)
; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG $vgpr0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret float %ret
}
declare float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32 immarg)
declare float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32, i32 immarg)
|
