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
212
213
214
215
216
|
;=========================== begin_copyright_notice ============================
;
; Copyright (C) 2017-2023 Intel Corporation
;
; SPDX-License-Identifier: MIT
;
;============================ end_copyright_notice =============================
; REQUIRES: regkeys
;
; RUN: igc_opt --typed-pointers %s -S -inputocl -igc-ldstcombine -regkey=EnableLdStCombine=5 \
; RUN: -platformbmg \
; RUN: | FileCheck %s
; CHECK-LABEL: target datalayout
; %__StructSOALayout_{{.*}} = type <{ %__StructAOSLayout_{{.*}} }>
; %__StructAOSLayout_{{.*}} = type <{ i8, i8, i8, i8 }>
;
target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v16:16:16-v24:32:32-v32:32:32-v48:64:64-v64:64:64-v96:128:128-v128:128:128-v192:256:256-v256:256:256-v512:512:512-v1024:1024:1024-n8:16:32"
target triple = "spir64-unknown-unknown"
; Function Attrs: convergent nounwind
define spir_kernel void @test_basic(i32 addrspace(1)* %d, i64 addrspace(1)* %dll,
i32 addrspace(1)* %si, i64 addrspace(1)* %sll, i16 %localIdX)
{
entry:
%conv.i.i = zext i16 %localIdX to i64
;
; case 0: load i8; load i8 --> load <2 x i8>
;
; CHECK-LABEL: define spir_kernel void @test_basic
; CHECK: load <2 x i8>
;
%c0.si = bitcast i32 addrspace(1)* %si to i8 addrspace(1)*
%c0.arrayidx = getelementptr inbounds i8, i8 addrspace(1)* %c0.si, i64 %conv.i.i
%c0.0 = load i8, i8 addrspace(1)* %c0.arrayidx, align 4
%c0.add.1 = add nuw nsw i64 %conv.i.i, 1
%c0.arrayidx.1 = getelementptr inbounds i8, i8 addrspace(1)* %c0.si, i64 %c0.add.1
%c0.1 = load i8, i8 addrspace(1)* %c0.arrayidx.1, align 1
%c0.0.0 = zext i8 %c0.0 to i16
%c0.1.0 = zext i8 %c0.1 to i16
%c0.1.1 = shl i16 %c0.1.0, 8
%c0.e0 = add i16 %c0.1.1, %c0.0.0
%c0.d = bitcast i32 addrspace(1)* %d to i16 addrspace(1)*
%c0.idx = add nuw nsw i64 %conv.i.i, 11
%c0.arrayidx1 = getelementptr inbounds i16, i16 addrspace(1)* %c0.d, i64 %c0.idx
store i16 %c0.e0, i16 addrspace(1)* %c0.arrayidx1, align 4
;
; case 1: load i8; load i8; load i8; load i8 --> load i32 --> bitcast to <4 x i8>
;
; CHECK-LABEL: store i16
; CHECK: [[T1:%.*]] = load i32
; CHECK: bitcast i32 [[T1]] to <4 x i8>
;
%c1.si = bitcast i32 addrspace(1)* %si to i8 addrspace(1)*
%c1.arrayidx = getelementptr inbounds i8, i8 addrspace(1)* %c1.si, i64 %conv.i.i
%c1.0 = load i8, i8 addrspace(1)* %c1.arrayidx, align 4
%c1.add.1= add nuw nsw i64 %conv.i.i, 1
%c1.arrayidx.1 = getelementptr inbounds i8, i8 addrspace(1)* %c1.si, i64 %c1.add.1
%c1.1 = load i8, i8 addrspace(1)* %c1.arrayidx.1, align 1
%c1.add.2 = add nuw nsw i64 %conv.i.i, 2
%c1.arrayidx.2 = getelementptr inbounds i8, i8 addrspace(1)* %c1.si, i64 %c1.add.2
%c1.2 = load i8, i8 addrspace(1)* %c1.arrayidx.2, align 1
%c1.add.3 = add nuw nsw i64 %conv.i.i, 3
%c1.arrayidx.3 = getelementptr inbounds i8, i8 addrspace(1)* %c1.si, i64 %c1.add.3
%c1.3 = load i8, i8 addrspace(1)* %c1.arrayidx.3, align 1
%c1.e0.0 = insertelement <4 x i8> undef, i8 %c1.0, i32 0
%c1.e0.1 = insertelement <4 x i8> %c1.e0.0, i8 %c1.1, i32 1
%c1.e0.2 = insertelement <4 x i8> %c1.e0.1, i8 %c1.2, i32 2
%c1.e0.3 = insertelement <4 x i8> %c1.e0.2, i8 %c1.3, i32 3
%c1.idx = add nuw nsw i64 %conv.i.i, 128
%c1.arrayidx1 = getelementptr inbounds i32, i32 addrspace(1)* %d, i64 %c1.idx
%c1.addr = bitcast i32 addrspace(1)* %c1.arrayidx1 to <4 x i8> addrspace(1)*
store <4 x i8> %c1.e0.2, <4 x i8> addrspace(1)* %c1.addr, align 4
;
; case 2: load i16; load i16 -> load i32 --> bitcast to <2 x i16>
;
; CHECK-LABEL: store <4 x i8>
; CHECK: [[T2:%.*]] = load i32
; CHECK: bitcast i32 [[T2]] to <2 x i16>
;
%c2.si = bitcast i32 addrspace(1)* %si to i16 addrspace(1)*
%c2.arrayidx = getelementptr inbounds i16, i16 addrspace(1)* %c2.si, i64 %conv.i.i
%c2.0 = load i16, i16 addrspace(1)* %c2.arrayidx, align 4
%c2.add.1 = add nuw nsw i64 %conv.i.i, 1
%c2.arrayidx.1 = getelementptr inbounds i16, i16 addrspace(1)* %c2.si, i64 %c2.add.1
%c2.1 = load i16, i16 addrspace(1)* %c2.arrayidx.1, align 2
%c2.0.0 = zext i16 %c2.0 to i32
%c2.1.0 = zext i16 %c2.1 to i32
%c2.1.1 = shl i32 %c2.1.0, 16
%c2.e0 = or i32 %c2.1.1, %c2.0.0
%c2.arrayidx1 = getelementptr inbounds i32, i32 addrspace(1)* %d, i64 %conv.i.i
store i32 %c2.e0, i32 addrspace(1)* %c2.arrayidx1, align 4
;
; case 3: load i32; load i32 -> load <2 x i32>
;
; CHECK-LABEL: store i32
; CHECK: load <2 x i32>
;
%c3.baseidx = add i64 %conv.i.i, 64
%c3.arrayidx = getelementptr inbounds i32, i32 addrspace(1)* %si, i64 %c3.baseidx
%c3.0 = load i32, i32 addrspace(1)* %c3.arrayidx, align 4
%c3.add.1= add nuw nsw i64 %c3.baseidx, 1
%c3.arrayidx.1 = getelementptr inbounds i32, i32 addrspace(1)* %si, i64 %c3.add.1
%c3.1 = load i32, i32 addrspace(1)* %c3.arrayidx.1, align 4
%c3.e0.0 = insertelement <2 x i32> undef, i32 %c3.0, i32 0
%c3.e0.1 = insertelement <2 x i32> %c3.e0.0, i32 %c3.1, i32 1
%c3.idx = add nuw nsw i64 %conv.i.i, 256
%c3.arrayidx1 = getelementptr inbounds i32, i32 addrspace(1)* %d, i64 %c3.idx
%c3.addr = bitcast i32 addrspace(1)* %c3.arrayidx1 to <2 x i32> addrspace(1)*
store <2 x i32> %c3.e0.1, <2 x i32> addrspace(1)* %c3.addr, align 4
;
; case 4: load i64; load i64 -> load <2 x i64>
;
; CHECK-LABEL: store <2 x i32>
; CHECK: load <2 x i64>
;
%c4.baseidx = add i64 %conv.i.i, 64
%c4.arrayidx = getelementptr inbounds i64, i64 addrspace(1)* %sll, i64 %c4.baseidx
%c4.0 = load i64, i64 addrspace(1)* %c4.arrayidx, align 8
%c4.add.1= add nuw nsw i64 %c4.baseidx, 1
%c4.arrayidx.1 = getelementptr inbounds i64, i64 addrspace(1)* %sll, i64 %c4.add.1
%c4.1 = load i64, i64 addrspace(1)* %c4.arrayidx.1, align 8
%c4.e0.0 = insertelement <2 x i64> undef, i64 %c4.0, i64 0
%c4.e0.1 = insertelement <2 x i64> %c4.e0.0, i64 %c4.1, i64 1
%c4.idx = add nuw nsw i64 %conv.i.i, 256
%c4.arrayidx1 = getelementptr inbounds i64, i64 addrspace(1)* %dll, i64 %c4.idx
%c4.addr = bitcast i64 addrspace(1)* %c4.arrayidx1 to <2 x i64> addrspace(1)*
store <2 x i64> %c4.e0.1, <2 x i64> addrspace(1)* %c4.addr, align 4
;
; case 5: load i32; load i32; load i64 -> load <4 x i32>
;
; CHECK-LABEL: store <2 x i64>
; CHECK: load <4 x i32>
;
%c5.baseidx = add i64 %conv.i.i, 512
%c5.arrayidx = getelementptr inbounds i32, i32 addrspace(1)* %si, i64 %c5.baseidx
%c5.0 = load i32, i32 addrspace(1)* %c5.arrayidx, align 4
%c5.add.1= add nuw nsw i64 %c5.baseidx, 1
%c5.arrayidx.1 = getelementptr inbounds i32, i32 addrspace(1)* %si, i64 %c5.add.1
%c5.1 = load i32, i32 addrspace(1)* %c5.arrayidx.1, align 8
%c5.add.2= add nuw nsw i64 %c5.baseidx, 2
%c5.arrayidx.2 = getelementptr inbounds i32, i32 addrspace(1)* %si, i64 %c5.add.2
%c5.arrayidx.2.0 = bitcast i32 addrspace(1)* %c5.arrayidx.2 to i64 addrspace(1)*
%c5.2 = load i64, i64 addrspace(1)* %c5.arrayidx.2.0, align 4
%c5.2.0 = trunc i64 %c5.2 to i32
%c5.2.1 = lshr i64 %c5.2, 32
%c5.2.2 = trunc i64 %c5.2.1 to i32
%c5.e0.0 = insertelement <4 x i32> undef, i32 %c5.0, i64 0
%c5.e0.1 = insertelement <4 x i32> %c5.e0.0, i32 %c5.1, i64 1
%c5.e0.2 = insertelement <4 x i32> %c5.e0.1, i32 %c5.2.0, i64 2
%c5.e0.3 = insertelement <4 x i32> %c5.e0.2, i32 %c5.2.2, i64 3
%c5.idx = add nuw nsw i64 %conv.i.i, 512
%c5.arrayidx1 = getelementptr inbounds i32, i32 addrspace(1)* %d, i64 %c5.idx
%c5.addr = bitcast i32 addrspace(1)* %c5.arrayidx1 to <4 x i32> addrspace(1)*
store <4 x i32> %c5.e0.3, <4 x i32> addrspace(1)* %c5.addr, align 4
;
; case 6: load i32 p+1; load i32 p -> load <2 x i32>
; This is to test that lead load is not the first and therefore, an address of
; of anchor load is used instead.
;
; CHECK-LABEL: c6.baseidx
; CHECK: %c6.arrayidx = getelementptr inbounds i32, i32 addrspace(1)* %si, i64 %c6.baseidx
; CHECK: [[T6_0:%.*]] = bitcast i32 addrspace(1)* %c6.arrayidx to i8 addrspace(1)*
; CHECK: [[T6_1:%.*]] = getelementptr inbounds i8, i8 addrspace(1)* [[T6_0]], i64 -4
; CHECK: [[T6_2:%.*]] = bitcast i8 addrspace(1)* [[T6_1]] to <2 x i32> addrspace(1)*
; CHECK: {{.*}} = load <2 x i32>, <2 x i32> addrspace(1)* [[T6_2]], align 4
;
%c6.baseidx = add i64 %conv.i.i, 577
%c6.arrayidx = getelementptr inbounds i32, i32 addrspace(1)* %si, i64 %c6.baseidx
%c6.0 = load i32, i32 addrspace(1)* %c6.arrayidx, align 4
%c6.baseidx.1 = add i64 %conv.i.i, 576
%c6.arrayidx.1 = getelementptr inbounds i32, i32 addrspace(1)* %si, i64 %c6.baseidx.1
%c6.1 = load i32, i32 addrspace(1)* %c6.arrayidx.1, align 4
%c6.v.0 = insertelement <2 x i32> undef, i32 %c6.0, i64 0
%c6.v.1 = insertelement <2 x i32> %c6.v.0, i32 %c6.1, i64 1
%c6.arrayidx1 = getelementptr inbounds i32, i32 addrspace(1)* %d, i64 %c6.baseidx
%c6.addr = bitcast i32 addrspace(1)* %c6.arrayidx1 to <2 x i32> addrspace(1)*
store <2 x i32> %c6.v.1, <2 x i32> addrspace(1)* %c6.addr, align 4
;
; case 7: load i8; load i8; load i16; --> load i32 --> bitcast to struct
;
; CHECK-LABEL: c7.baseidx
; CHECK: [[T7:%.*]] = load i32
; CHECK: call %__StructSOALayout_{{.*}} @llvm.genx.GenISA.bitcasttostruct.__StructSOALayout_{{.*}}.i32(i32 [[T7]])
;
%c7.baseidx = add i64 %conv.i.i, 1000
%c7.si = bitcast i32 addrspace(1)* %si to i8 addrspace(1)*
%c7.arrayidx = getelementptr inbounds i8, i8 addrspace(1)* %c7.si, i64 %c7.baseidx
%c7.0 = load i8, i8 addrspace(1)* %c7.arrayidx, align 4
%c7.add.1= add nuw nsw i64 %c7.baseidx, 1
%c7.arrayidx.1 = getelementptr inbounds i8, i8 addrspace(1)* %c7.si, i64 %c7.add.1
%c7.1 = load i8, i8 addrspace(1)* %c7.arrayidx.1, align 1
%c7.add.2 = add nuw nsw i64 %c7.baseidx, 2
%c7.arrayidx.2 = getelementptr inbounds i8, i8 addrspace(1)* %c7.si, i64 %c7.add.2
%c7.arrayidx.3 = bitcast i8 addrspace(1)* %c7.arrayidx.2 to i16 addrspace(1)*
%c7.2 = load i16, i16 addrspace(1)* %c7.arrayidx.3, align 2
%c7.2.0 = bitcast i16 %c7.2 to <2 x i8>
%c7.2.1 = extractelement <2 x i8> %c7.2.0, i32 0
%c7.2.2 = extractelement <2 x i8> %c7.2.0, i32 1
%c7.e0.0 = insertelement <4 x i8> undef, i8 %c7.0, i32 0
%c7.e0.1 = insertelement <4 x i8> %c7.e0.0, i8 %c7.1, i32 1
%c7.e0.2 = insertelement <4 x i8> %c7.e0.1, i8 %c7.2.1, i32 2
%c7.e0.3 = insertelement <4 x i8> %c7.e0.2, i8 %c7.2.2, i32 3
%c7.arrayidx1 = getelementptr inbounds i32, i32 addrspace(1)* %d, i64 %c7.baseidx
%c7.addr = bitcast i32 addrspace(1)* %c7.arrayidx1 to <4 x i8> addrspace(1)*
store <4 x i8> %c7.e0.2, <4 x i8> addrspace(1)* %c7.addr, align 4
; CHECK-LABEL: ret void
ret void
}
|
