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
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=amdgcn -mcpu=verde < %s | FileCheck -enable-var-scope -check-prefixes=GCN %s
define amdgpu_kernel void @sext_i16_to_i32_uniform(ptr addrspace(1) %out, i16 %a, i32 %b) {
; GCN-LABEL: sext_i16_to_i32_uniform:
; GCN: ; %bb.0:
; GCN-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x9
; GCN-NEXT: s_mov_b32 s7, 0xf000
; GCN-NEXT: s_mov_b32 s6, -1
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: s_mov_b32 s4, s0
; GCN-NEXT: s_sext_i32_i16 s0, s2
; GCN-NEXT: s_add_i32 s0, s3, s0
; GCN-NEXT: s_mov_b32 s5, s1
; GCN-NEXT: v_mov_b32_e32 v0, s0
; GCN-NEXT: buffer_store_dword v0, off, s[4:7], 0
; GCN-NEXT: s_endpgm
%sext = sext i16 %a to i32
%res = add i32 %b, %sext
store i32 %res, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @sext_i16_to_i64_uniform(ptr addrspace(1) %out, i16 %a, i64 %b) {
; GCN-LABEL: sext_i16_to_i64_uniform:
; GCN: ; %bb.0:
; GCN-NEXT: s_load_dword s6, s[4:5], 0xb
; GCN-NEXT: s_load_dwordx2 s[8:9], s[4:5], 0xd
; GCN-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x9
; GCN-NEXT: s_mov_b32 s3, 0xf000
; GCN-NEXT: s_mov_b32 s2, -1
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: s_bfe_i64 s[4:5], s[6:7], 0x100000
; GCN-NEXT: s_add_u32 s4, s8, s4
; GCN-NEXT: s_addc_u32 s5, s9, s5
; GCN-NEXT: v_mov_b32_e32 v0, s4
; GCN-NEXT: v_mov_b32_e32 v1, s5
; GCN-NEXT: buffer_store_dwordx2 v[0:1], off, s[0:3], 0
; GCN-NEXT: s_endpgm
%sext = sext i16 %a to i64
%res = add i64 %b, %sext
store i64 %res, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @sext_i16_to_i32_divergent(ptr addrspace(1) %out, i16 %a, i32 %b) {
; GCN-LABEL: sext_i16_to_i32_divergent:
; GCN: ; %bb.0:
; GCN-NEXT: s_load_dword s6, s[4:5], 0xb
; GCN-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x9
; GCN-NEXT: s_mov_b32 s3, 0xf000
; GCN-NEXT: s_mov_b32 s2, -1
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: v_add_i32_e32 v0, vcc, s6, v0
; GCN-NEXT: v_bfe_i32 v0, v0, 0, 16
; GCN-NEXT: buffer_store_dword v0, off, s[0:3], 0
; GCN-NEXT: s_endpgm
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.truncated = trunc i32 %tid to i16
%divergent.a = add i16 %a, %tid.truncated
%sext = sext i16 %divergent.a to i32
store i32 %sext, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @sext_i16_to_i64_divergent(ptr addrspace(1) %out, i16 %a, i64 %b) {
; GCN-LABEL: sext_i16_to_i64_divergent:
; GCN: ; %bb.0:
; GCN-NEXT: s_load_dword s6, s[4:5], 0xb
; GCN-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x9
; GCN-NEXT: s_mov_b32 s3, 0xf000
; GCN-NEXT: s_mov_b32 s2, -1
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: v_add_i32_e32 v0, vcc, s6, v0
; GCN-NEXT: v_bfe_i32 v0, v0, 0, 16
; GCN-NEXT: v_ashrrev_i32_e32 v1, 31, v0
; GCN-NEXT: buffer_store_dwordx2 v[0:1], off, s[0:3], 0
; GCN-NEXT: s_endpgm
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.truncated = trunc i32 %tid to i16
%divergent.a = add i16 %a, %tid.truncated
%sext = sext i16 %divergent.a to i64
store i64 %sext, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @sext_i32_to_i64_uniform(ptr addrspace(1) %out, i32 %a, i64 %b) {
; GCN-LABEL: sext_i32_to_i64_uniform:
; GCN: ; %bb.0:
; GCN-NEXT: s_load_dword s8, s[4:5], 0xb
; GCN-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0xd
; GCN-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x9
; GCN-NEXT: s_mov_b32 s3, 0xf000
; GCN-NEXT: s_mov_b32 s2, -1
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: s_ashr_i32 s5, s8, 31
; GCN-NEXT: s_add_u32 s4, s6, s8
; GCN-NEXT: s_addc_u32 s5, s7, s5
; GCN-NEXT: v_mov_b32_e32 v0, s4
; GCN-NEXT: v_mov_b32_e32 v1, s5
; GCN-NEXT: buffer_store_dwordx2 v[0:1], off, s[0:3], 0
; GCN-NEXT: s_endpgm
%sext = sext i32 %a to i64
%res = add i64 %b, %sext
store i64 %res, ptr addrspace(1) %out
ret void
}
define amdgpu_kernel void @sext_i32_to_i64_divergent(ptr addrspace(1) %out, i32 %a, i64 %b) {
; GCN-LABEL: sext_i32_to_i64_divergent:
; GCN: ; %bb.0:
; GCN-NEXT: s_load_dword s6, s[4:5], 0xb
; GCN-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x9
; GCN-NEXT: s_mov_b32 s3, 0xf000
; GCN-NEXT: s_mov_b32 s2, -1
; GCN-NEXT: s_waitcnt lgkmcnt(0)
; GCN-NEXT: v_add_i32_e32 v0, vcc, s6, v0
; GCN-NEXT: v_ashrrev_i32_e32 v1, 31, v0
; GCN-NEXT: buffer_store_dwordx2 v[0:1], off, s[0:3], 0
; GCN-NEXT: s_endpgm
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%divergent.a = add i32 %a, %tid
%sext = sext i32 %divergent.a to i64
store i64 %sext, ptr addrspace(1) %out
ret void
}
declare i32 @llvm.amdgcn.workitem.id.x() #1
attributes #0 = { nounwind }
attributes #1 = { nounwind readnone speculatable }
|
