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
|
; RUN: llc -mtriple amdgcn--amdhsa -mcpu=fiji -amdgpu-scalarize-global-loads=true < %s | FileCheck %s
; uniform loads
; CHECK-LABEL: @uniform_load
; CHECK: s_load_dwordx4
; CHECK-NOT: flat_load_dword
define amdgpu_kernel void @uniform_load(ptr addrspace(1) %arg, [8 x i32], ptr addrspace(1) %arg1) {
bb:
%tmp2 = load float, ptr addrspace(1) %arg, align 4, !tbaa !8
%tmp3 = fadd float %tmp2, 0.000000e+00
%tmp4 = getelementptr inbounds float, ptr addrspace(1) %arg, i64 1
%tmp5 = load float, ptr addrspace(1) %tmp4, align 4, !tbaa !8
%tmp6 = fadd float %tmp3, %tmp5
%tmp7 = getelementptr inbounds float, ptr addrspace(1) %arg, i64 2
%tmp8 = load float, ptr addrspace(1) %tmp7, align 4, !tbaa !8
%tmp9 = fadd float %tmp6, %tmp8
%tmp10 = getelementptr inbounds float, ptr addrspace(1) %arg, i64 3
%tmp11 = load float, ptr addrspace(1) %tmp10, align 4, !tbaa !8
%tmp12 = fadd float %tmp9, %tmp11
%tmp13 = getelementptr inbounds float, ptr addrspace(1) %arg1
store float %tmp12, ptr addrspace(1) %tmp13, align 4, !tbaa !8
ret void
}
; uniform loads before and after an aliasing store
; CHECK-LABEL: @uniform_load_store_load
; CHECK: s_load_dwordx4
; CHECK: s_load_dword
; CHECK: flat_store_dword
; CHECK: flat_load_dword
; CHECK: flat_store_dword
define amdgpu_kernel void @uniform_load_store_load(ptr addrspace(1) %arg0, ptr addrspace(1) %arg1) {
bb:
%tmp2 = load float, ptr addrspace(1) %arg0, !tbaa !8
store float %tmp2, ptr addrspace(1) %arg1, !tbaa !8
%tmp3 = load float, ptr addrspace(1) %arg0, !tbaa !8
store float %tmp3, ptr addrspace(1) %arg1, !tbaa !8
ret void
}
; non-uniform loads
; CHECK-LABEL: @non-uniform_load
; CHECK: flat_load_dword
; CHECK-NOT: s_load_dwordx4
define amdgpu_kernel void @non-uniform_load(ptr addrspace(1) %arg, [8 x i32], ptr addrspace(1) %arg1) #0 {
bb:
%tmp = call i32 @llvm.amdgcn.workitem.id.x() #1
%tmp2 = getelementptr inbounds float, ptr addrspace(1) %arg, i32 %tmp
%tmp3 = load float, ptr addrspace(1) %tmp2, align 4, !tbaa !8
%tmp4 = fadd float %tmp3, 0.000000e+00
%tmp5 = add i32 %tmp, 1
%tmp6 = getelementptr inbounds float, ptr addrspace(1) %arg, i32 %tmp5
%tmp7 = load float, ptr addrspace(1) %tmp6, align 4, !tbaa !8
%tmp8 = fadd float %tmp4, %tmp7
%tmp9 = add i32 %tmp, 2
%tmp10 = getelementptr inbounds float, ptr addrspace(1) %arg, i32 %tmp9
%tmp11 = load float, ptr addrspace(1) %tmp10, align 4, !tbaa !8
%tmp12 = fadd float %tmp8, %tmp11
%tmp13 = add i32 %tmp, 3
%tmp14 = getelementptr inbounds float, ptr addrspace(1) %arg, i32 %tmp13
%tmp15 = load float, ptr addrspace(1) %tmp14, align 4, !tbaa !8
%tmp16 = fadd float %tmp12, %tmp15
%tmp17 = getelementptr inbounds float, ptr addrspace(1) %arg1, i32 %tmp
store float %tmp16, ptr addrspace(1) %tmp17, align 4, !tbaa !8
ret void
}
; uniform load dominated by no-alias store - scalarize
; CHECK-LABEL: @no_memdep_alias_arg
; CHECK: s_load_dwordx2 s[[[IN_LO:[0-9]+]]:[[IN_HI:[0-9]+]]], s[8:9], 0x0
; CHECK: s_load_dword [[SVAL:s[0-9]+]], s[[[IN_LO]]:[[IN_HI]]], 0x0
; CHECK: v_mov_b32_e32 [[VVAL:v[0-9]+]], [[SVAL]]
; CHECK: flat_store_dword v[{{[0-9]+:[0-9]+}}], [[VVAL]]
define amdgpu_kernel void @no_memdep_alias_arg(ptr addrspace(1) noalias %in, [8 x i32], ptr addrspace(1) %out0, [8 x i32], ptr addrspace(1) %out1) {
store i32 0, ptr addrspace(1) %out0
%val = load i32, ptr addrspace(1) %in
store i32 %val, ptr addrspace(1) %out1
ret void
}
; uniform load dominated by alias store - vector
; CHECK-LABEL: {{^}}memdep:
; CHECK: flat_store_dword
; CHECK: flat_load_dword [[VVAL:v[0-9]+]]
; CHECK: flat_store_dword v[{{[0-9]+:[0-9]+}}], [[VVAL]]
define amdgpu_kernel void @memdep(ptr addrspace(1) %in, [8 x i32], ptr addrspace(1) %out0, [8 x i32], ptr addrspace(1) %out1) {
store i32 0, ptr addrspace(1) %out0
%val = load i32, ptr addrspace(1) %in
store i32 %val, ptr addrspace(1) %out1
ret void
}
; uniform load from global array
; CHECK-LABEL: @global_array
; CHECK: s_getpc_b64 [[GET_PC:s\[[0-9]+:[0-9]+\]]]
; CHECK-DAG: s_load_dwordx2 [[A_ADDR:s\[[0-9]+:[0-9]+\]]], [[GET_PC]], 0x0
; CHECK-DAG: s_load_dwordx2 [[A_ADDR1:s\[[0-9]+:[0-9]+\]]], [[A_ADDR]], 0x0
; CHECK-DAG: s_load_dwordx2 [[OUT:s\[[0-9]+:[0-9]+\]]], s[8:9], 0x0
; CHECK-DAG: s_load_dword [[SVAL:s[0-9]+]], [[A_ADDR1]], 0x0
; CHECK: v_mov_b32_e32 [[VVAL:v[0-9]+]], [[SVAL]]
; CHECK: flat_store_dword v[{{[0-9]+:[0-9]+}}], [[VVAL]]
@A = common local_unnamed_addr addrspace(1) global ptr addrspace(1) null, align 4
define amdgpu_kernel void @global_array(ptr addrspace(1) nocapture %out) {
entry:
%load0 = load ptr addrspace(1), ptr addrspace(1) @A, align 4
%load1 = load i32, ptr addrspace(1) %load0, align 4
store i32 %load1, ptr addrspace(1) %out, align 4
ret void
}
; uniform load from global array dominated by alias store
; CHECK-LABEL: @global_array_alias_store
; CHECK: flat_store_dword
; CHECK: v_mov_b32_e32 v[[ADDR_LO:[0-9]+]], s{{[0-9]+}}
; CHECK: v_mov_b32_e32 v[[ADDR_HI:[0-9]+]], s{{[0-9]+}}
; CHECK: flat_load_dwordx2 [[A_ADDR:v\[[0-9]+:[0-9]+\]]], v[[[ADDR_LO]]:[[ADDR_HI]]]
; CHECK: flat_load_dword [[VVAL:v[0-9]+]], [[A_ADDR]]
; CHECK: flat_store_dword v[{{[0-9]+:[0-9]+}}], [[VVAL]]
define amdgpu_kernel void @global_array_alias_store(ptr addrspace(1) nocapture %out, [8 x i32], i32 %n) {
entry:
%gep = getelementptr i32, ptr addrspace(1) %out, i32 %n
store i32 12, ptr addrspace(1) %gep
%load0 = load ptr addrspace(1), ptr addrspace(1) @A, align 4
%load1 = load i32, ptr addrspace(1) %load0, align 4
store i32 %load1, ptr addrspace(1) %out, align 4
ret void
}
declare i32 @llvm.amdgcn.workitem.id.x() #1
attributes #1 = { nounwind readnone }
!8 = !{!9, !9, i64 0}
!9 = !{!"float", !10, i64 0}
!10 = !{!"omnipotent char", !11, i64 0}
!11 = !{!"Simple C/C++ TBAA"}
|
