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
|
; RUN: llc -march=amdgcn -mcpu=tahiti < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
; RUN: llc -march=amdgcn -mcpu=tonga < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
; FUNC-LABEL: {{^}}test_copy_v4i8:
; SI: buffer_load_dword [[REG:v[0-9]+]]
; SI: buffer_store_dword [[REG]]
; SI: s_endpgm
define void @test_copy_v4i8(<4 x i8> addrspace(1)* %out, <4 x i8> addrspace(1)* %in) nounwind {
%val = load <4 x i8>, <4 x i8> addrspace(1)* %in, align 4
store <4 x i8> %val, <4 x i8> addrspace(1)* %out, align 4
ret void
}
; FUNC-LABEL: {{^}}test_copy_v4i8_x2:
; SI: buffer_load_dword [[REG:v[0-9]+]]
; SI: buffer_store_dword [[REG]]
; SI: buffer_store_dword [[REG]]
; SI: s_endpgm
define void @test_copy_v4i8_x2(<4 x i8> addrspace(1)* %out0, <4 x i8> addrspace(1)* %out1, <4 x i8> addrspace(1)* %in) nounwind {
%val = load <4 x i8>, <4 x i8> addrspace(1)* %in, align 4
store <4 x i8> %val, <4 x i8> addrspace(1)* %out0, align 4
store <4 x i8> %val, <4 x i8> addrspace(1)* %out1, align 4
ret void
}
; FUNC-LABEL: {{^}}test_copy_v4i8_x3:
; SI: buffer_load_dword [[REG:v[0-9]+]]
; SI: buffer_store_dword [[REG]]
; SI: buffer_store_dword [[REG]]
; SI: buffer_store_dword [[REG]]
; SI: s_endpgm
define void @test_copy_v4i8_x3(<4 x i8> addrspace(1)* %out0, <4 x i8> addrspace(1)* %out1, <4 x i8> addrspace(1)* %out2, <4 x i8> addrspace(1)* %in) nounwind {
%val = load <4 x i8>, <4 x i8> addrspace(1)* %in, align 4
store <4 x i8> %val, <4 x i8> addrspace(1)* %out0, align 4
store <4 x i8> %val, <4 x i8> addrspace(1)* %out1, align 4
store <4 x i8> %val, <4 x i8> addrspace(1)* %out2, align 4
ret void
}
; FUNC-LABEL: {{^}}test_copy_v4i8_x4:
; SI: buffer_load_dword [[REG:v[0-9]+]]
; SI: buffer_store_dword [[REG]]
; SI: buffer_store_dword [[REG]]
; SI: buffer_store_dword [[REG]]
; SI: buffer_store_dword [[REG]]
; SI: s_endpgm
define void @test_copy_v4i8_x4(<4 x i8> addrspace(1)* %out0, <4 x i8> addrspace(1)* %out1, <4 x i8> addrspace(1)* %out2, <4 x i8> addrspace(1)* %out3, <4 x i8> addrspace(1)* %in) nounwind {
%val = load <4 x i8>, <4 x i8> addrspace(1)* %in, align 4
store <4 x i8> %val, <4 x i8> addrspace(1)* %out0, align 4
store <4 x i8> %val, <4 x i8> addrspace(1)* %out1, align 4
store <4 x i8> %val, <4 x i8> addrspace(1)* %out2, align 4
store <4 x i8> %val, <4 x i8> addrspace(1)* %out3, align 4
ret void
}
; FUNC-LABEL: {{^}}test_copy_v4i8_extra_use:
; SI: buffer_load_ubyte
; SI: buffer_load_ubyte
; SI: buffer_load_ubyte
; SI: buffer_load_ubyte
; SI-DAG: v_add
; SI-DAG: v_add
; SI-DAG: v_add
; SI-DAG: v_add
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI_DAG: buffer_store_byte
; After scalarizing v4i8 loads is fixed.
; XSI: buffer_load_dword
; XSI: V_BFE
; XSI: V_ADD
; XSI: V_ADD
; XSI: V_ADD
; XSI: buffer_store_dword
; XSI: buffer_store_dword
; SI: s_endpgm
define void @test_copy_v4i8_extra_use(<4 x i8> addrspace(1)* %out0, <4 x i8> addrspace(1)* %out1, <4 x i8> addrspace(1)* %in) nounwind {
%val = load <4 x i8>, <4 x i8> addrspace(1)* %in, align 4
%add = add <4 x i8> %val, <i8 9, i8 9, i8 9, i8 9>
store <4 x i8> %val, <4 x i8> addrspace(1)* %out0, align 4
store <4 x i8> %add, <4 x i8> addrspace(1)* %out1, align 4
ret void
}
; FUNC-LABEL: {{^}}test_copy_v4i8_x2_extra_use:
; SI: buffer_load_ubyte
; SI: buffer_load_ubyte
; SI: buffer_load_ubyte
; SI: buffer_load_ubyte
; SI-DAG: v_add
; SI-DAG: v_add
; SI-DAG: v_add
; SI-DAG: v_add
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI_DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI-DAG: buffer_store_byte
; SI_DAG: buffer_store_byte
; XSI: buffer_load_dword
; XSI: BFE
; XSI: buffer_store_dword
; XSI: V_ADD
; XSI: buffer_store_dword
; XSI-NEXT: buffer_store_dword
; SI: s_endpgm
define void @test_copy_v4i8_x2_extra_use(<4 x i8> addrspace(1)* %out0, <4 x i8> addrspace(1)* %out1, <4 x i8> addrspace(1)* %out2, <4 x i8> addrspace(1)* %in) nounwind {
%val = load <4 x i8>, <4 x i8> addrspace(1)* %in, align 4
%add = add <4 x i8> %val, <i8 9, i8 9, i8 9, i8 9>
store <4 x i8> %val, <4 x i8> addrspace(1)* %out0, align 4
store <4 x i8> %add, <4 x i8> addrspace(1)* %out1, align 4
store <4 x i8> %val, <4 x i8> addrspace(1)* %out2, align 4
ret void
}
; FUNC-LABEL: {{^}}test_copy_v3i8:
; SI-NOT: bfe
; SI-NOT: bfi
; SI: s_endpgm
define void @test_copy_v3i8(<3 x i8> addrspace(1)* %out, <3 x i8> addrspace(1)* %in) nounwind {
%val = load <3 x i8>, <3 x i8> addrspace(1)* %in, align 4
store <3 x i8> %val, <3 x i8> addrspace(1)* %out, align 4
ret void
}
; FUNC-LABEL: {{^}}test_copy_v4i8_volatile_load:
; SI: buffer_load_ubyte
; SI: buffer_load_ubyte
; SI: buffer_load_ubyte
; SI: buffer_load_ubyte
; SI: s_endpgm
define void @test_copy_v4i8_volatile_load(<4 x i8> addrspace(1)* %out, <4 x i8> addrspace(1)* %in) nounwind {
%val = load volatile <4 x i8>, <4 x i8> addrspace(1)* %in, align 4
store <4 x i8> %val, <4 x i8> addrspace(1)* %out, align 4
ret void
}
; FUNC-LABEL: {{^}}test_copy_v4i8_volatile_store:
; SI: buffer_load_ubyte
; SI: buffer_load_ubyte
; SI: buffer_load_ubyte
; SI: buffer_load_ubyte
; SI: buffer_store_byte
; SI: buffer_store_byte
; SI: buffer_store_byte
; SI: buffer_store_byte
; SI: s_endpgm
define void @test_copy_v4i8_volatile_store(<4 x i8> addrspace(1)* %out, <4 x i8> addrspace(1)* %in) nounwind {
%val = load <4 x i8>, <4 x i8> addrspace(1)* %in, align 4
store volatile <4 x i8> %val, <4 x i8> addrspace(1)* %out, align 4
ret void
}
|
