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
|
; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-4,-unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT4,ELT4-ALIGNED,ALIGNED,ALL %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-8,-unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT8,ALIGNED,ALL %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-16,-unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT16,ALIGNED,ALL %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-4,+unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT4,ELT4-UNALIGNED,UNALIGNED,ALL %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-8,+unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT8,ELT8-UNALIGNED,UNALIGNED,ALL %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-16,+unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT16,ELT16-UNALIGNED,UNALIGNED,ALL %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v4i32
; ELT4-ALIGNED: store i32
; ELT4-ALIGNED: store i32
; ELT4-ALIGNED: store i32
; ELT4-ALIGNED: store i32
; ELT8: store <2 x i32>
; ELT8: store <2 x i32>
; ELT16-UNALIGNED: store <4 x i32>
define amdgpu_kernel void @merge_private_store_4_vector_elts_loads_v4i32(ptr addrspace(5) %out) #0 {
%out.gep.1 = getelementptr i32, ptr addrspace(5) %out, i32 1
%out.gep.2 = getelementptr i32, ptr addrspace(5) %out, i32 2
%out.gep.3 = getelementptr i32, ptr addrspace(5) %out, i32 3
store i32 9, ptr addrspace(5) %out
store i32 1, ptr addrspace(5) %out.gep.1
store i32 23, ptr addrspace(5) %out.gep.2
store i32 19, ptr addrspace(5) %out.gep.3
ret void
}
; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v4i32_align1(
; ALIGNED: store i32 9, ptr addrspace(5) %out, align 1
; ALIGNED: store i32 1, ptr addrspace(5) %out.gep.1, align 1
; ALIGNED: store i32 23, ptr addrspace(5) %out.gep.2, align 1
; ALIGNED: store i32 19, ptr addrspace(5) %out.gep.3, align 1
; ELT16-UNALIGNED: store <4 x i32> <i32 9, i32 1, i32 23, i32 19>, ptr addrspace(5) %out, align 1
; ELT8-UNALIGNED: store <2 x i32> <i32 9, i32 1>, ptr addrspace(5) %out, align 1
; ELT8-UNALIGNED: store <2 x i32> <i32 23, i32 19>, ptr addrspace(5) %out.gep.2, align 1
; ELT4-UNALIGNED: store i32
; ELT4-UNALIGNED: store i32
; ELT4-UNALIGNED: store i32
; ELT4-UNALIGNED: store i32
define amdgpu_kernel void @merge_private_store_4_vector_elts_loads_v4i32_align1(ptr addrspace(5) %out) #0 {
%out.gep.1 = getelementptr i32, ptr addrspace(5) %out, i32 1
%out.gep.2 = getelementptr i32, ptr addrspace(5) %out, i32 2
%out.gep.3 = getelementptr i32, ptr addrspace(5) %out, i32 3
store i32 9, ptr addrspace(5) %out, align 1
store i32 1, ptr addrspace(5) %out.gep.1, align 1
store i32 23, ptr addrspace(5) %out.gep.2, align 1
store i32 19, ptr addrspace(5) %out.gep.3, align 1
ret void
}
; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v4i32_align2(
; ALL: store i32
; ALL: store i32
; ALL: store i32
; ALL: store i32
define amdgpu_kernel void @merge_private_store_4_vector_elts_loads_v4i32_align2(ptr addrspace(5) %out) #0 {
%out.gep.1 = getelementptr i32, ptr addrspace(5) %out, i32 1
%out.gep.2 = getelementptr i32, ptr addrspace(5) %out, i32 2
%out.gep.3 = getelementptr i32, ptr addrspace(5) %out, i32 3
store i32 9, ptr addrspace(5) %out, align 2
store i32 1, ptr addrspace(5) %out.gep.1, align 2
store i32 23, ptr addrspace(5) %out.gep.2, align 2
store i32 19, ptr addrspace(5) %out.gep.3, align 2
ret void
}
; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v4i8(
; ALL: store <4 x i8>
define amdgpu_kernel void @merge_private_store_4_vector_elts_loads_v4i8(ptr addrspace(5) %out) #0 {
%out.gep.1 = getelementptr i8, ptr addrspace(5) %out, i32 1
%out.gep.2 = getelementptr i8, ptr addrspace(5) %out, i32 2
%out.gep.3 = getelementptr i8, ptr addrspace(5) %out, i32 3
store i8 9, ptr addrspace(5) %out, align 4
store i8 1, ptr addrspace(5) %out.gep.1
store i8 23, ptr addrspace(5) %out.gep.2
store i8 19, ptr addrspace(5) %out.gep.3
ret void
}
; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v4i8_align1(
; ALIGNED: store i8
; ALIGNED: store i8
; ALIGNED: store i8
; ALIGNED: store i8
; UNALIGNED: store <4 x i8> <i8 9, i8 1, i8 23, i8 19>, ptr addrspace(5) %out, align 1
define amdgpu_kernel void @merge_private_store_4_vector_elts_loads_v4i8_align1(ptr addrspace(5) %out) #0 {
%out.gep.1 = getelementptr i8, ptr addrspace(5) %out, i32 1
%out.gep.2 = getelementptr i8, ptr addrspace(5) %out, i32 2
%out.gep.3 = getelementptr i8, ptr addrspace(5) %out, i32 3
store i8 9, ptr addrspace(5) %out, align 1
store i8 1, ptr addrspace(5) %out.gep.1, align 1
store i8 23, ptr addrspace(5) %out.gep.2, align 1
store i8 19, ptr addrspace(5) %out.gep.3, align 1
ret void
}
; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v2i16(
; ALL: store <2 x i16>
define amdgpu_kernel void @merge_private_store_4_vector_elts_loads_v2i16(ptr addrspace(5) %out) #0 {
%out.gep.1 = getelementptr i16, ptr addrspace(5) %out, i32 1
store i16 9, ptr addrspace(5) %out, align 4
store i16 12, ptr addrspace(5) %out.gep.1
ret void
}
; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v2i16_align2(
; ALL: store i16
; ALL: store i16
define amdgpu_kernel void @merge_private_store_4_vector_elts_loads_v2i16_align2(ptr addrspace(5) %out) #0 {
%out.gep.1 = getelementptr i16, ptr addrspace(5) %out, i32 1
store i16 9, ptr addrspace(5) %out, align 2
store i16 12, ptr addrspace(5) %out.gep.1, align 2
ret void
}
; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v2i16_align1(
; ALIGNED: store i16
; ALIGNED: store i16
; UNALIGNED: store <2 x i16> <i16 9, i16 12>, ptr addrspace(5) %out, align 1
define amdgpu_kernel void @merge_private_store_4_vector_elts_loads_v2i16_align1(ptr addrspace(5) %out) #0 {
%out.gep.1 = getelementptr i16, ptr addrspace(5) %out, i32 1
store i16 9, ptr addrspace(5) %out, align 1
store i16 12, ptr addrspace(5) %out.gep.1, align 1
ret void
}
; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v2i16_align8(
; ALL: store <2 x i16> <i16 9, i16 12>, ptr addrspace(5) %out, align 8
define amdgpu_kernel void @merge_private_store_4_vector_elts_loads_v2i16_align8(ptr addrspace(5) %out) #0 {
%out.gep.1 = getelementptr i16, ptr addrspace(5) %out, i32 1
store i16 9, ptr addrspace(5) %out, align 8
store i16 12, ptr addrspace(5) %out.gep.1, align 2
ret void
}
; ALL-LABEL: @merge_private_store_3_vector_elts_loads_v4i32
; ELT4: store i32
; ELT4: store i32
; ELT4: store i32
; ELT8: store <2 x i32>
; ELT8: store i32
; ELT16: store <3 x i32>
define amdgpu_kernel void @merge_private_store_3_vector_elts_loads_v4i32(ptr addrspace(5) %out) #0 {
%out.gep.1 = getelementptr i32, ptr addrspace(5) %out, i32 1
%out.gep.2 = getelementptr i32, ptr addrspace(5) %out, i32 2
store i32 9, ptr addrspace(5) %out
store i32 1, ptr addrspace(5) %out.gep.1
store i32 23, ptr addrspace(5) %out.gep.2
ret void
}
; ALL-LABEL: @merge_private_store_3_vector_elts_loads_v4i32_align1(
; ALIGNED: store i32
; ALIGNED: store i32
; ALIGNED: store i32
; ELT4-UNALIGNED: store i32
; ELT4-UNALIGNED: store i32
; ELT4-UNALIGNED: store i32
; ELT8-UNALIGNED: store <2 x i32>
; ELT8-UNALIGNED: store i32
; ELT16-UNALIGNED: store <3 x i32>
define amdgpu_kernel void @merge_private_store_3_vector_elts_loads_v4i32_align1(ptr addrspace(5) %out) #0 {
%out.gep.1 = getelementptr i32, ptr addrspace(5) %out, i32 1
%out.gep.2 = getelementptr i32, ptr addrspace(5) %out, i32 2
store i32 9, ptr addrspace(5) %out, align 1
store i32 1, ptr addrspace(5) %out.gep.1, align 1
store i32 23, ptr addrspace(5) %out.gep.2, align 1
ret void
}
; ALL-LABEL: @merge_private_store_3_vector_elts_loads_v4i8_align1(
; ALIGNED: store i8
; ALIGNED: store i8
; ALIGNED: store i8
; UNALIGNED: store <3 x i8>
define amdgpu_kernel void @merge_private_store_3_vector_elts_loads_v4i8_align1(ptr addrspace(5) %out) #0 {
%out.gep.1 = getelementptr i8, ptr addrspace(5) %out, i8 1
%out.gep.2 = getelementptr i8, ptr addrspace(5) %out, i8 2
store i8 9, ptr addrspace(5) %out, align 1
store i8 1, ptr addrspace(5) %out.gep.1, align 1
store i8 23, ptr addrspace(5) %out.gep.2, align 1
ret void
}
attributes #0 = { nounwind }
|
