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
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
|
; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 5
; RUN: opt -passes='print<access-info>' -disable-output -mtriple=riscv64 -mattr=+v < %s 2>&1 | FileCheck %s
; RUN: opt -passes='print<access-info>' -disable-output -mtriple=x86_64 < %s 2>&1 | FileCheck %s
; REQUIRES: riscv-registered-target, x86-registered-target
; Dependence distance between read and write is greater than the trip
; count of the loop. Thus, values written are never read for any
; valid vectorization of the loop.
define void @test(ptr %p) {
; CHECK-LABEL: 'test'
; CHECK-NEXT: loop:
; CHECK-NEXT: Memory dependences are safe
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv = phi i64 [0, %entry], [%iv.next, %loop]
%a1 = getelementptr i64, ptr %p, i64 %iv
%v = load i64, ptr %a1, align 8
%offset = add i64 %iv, 200
%a2 = getelementptr i64, ptr %p, i64 %offset
store i64 %v, ptr %a2, align 8
%iv.next = add i64 %iv, 1
%cmp = icmp ne i64 %iv, 199
br i1 %cmp, label %loop, label %exit
exit:
ret void
}
; Dependence distance is less than trip count, thus we must prove that
; chosen VF guaranteed to be less than dependence distance.
define void @test_may_clobber1(ptr %p) {
; CHECK-LABEL: 'test_may_clobber1'
; CHECK-NEXT: loop:
; CHECK-NEXT: Memory dependences are safe with a maximum safe vector width of 6400 bits, with a maximum safe store-load forward width of 256 bits
; CHECK-NEXT: Dependences:
; CHECK-NEXT: BackwardVectorizable:
; CHECK-NEXT: %v = load i64, ptr %a1, align 32 ->
; CHECK-NEXT: store i64 %v, ptr %a2, align 32
; CHECK-EMPTY:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv = phi i64 [0, %entry], [%iv.next, %loop]
%a1 = getelementptr i64, ptr %p, i64 %iv
%v = load i64, ptr %a1, align 32
%offset = add i64 %iv, 100
%a2 = getelementptr i64, ptr %p, i64 %offset
store i64 %v, ptr %a2, align 32
%iv.next = add i64 %iv, 1
%cmp = icmp ne i64 %iv, 199
br i1 %cmp, label %loop, label %exit
exit:
ret void
}
define void @test_may_clobber2(ptr %p) {
; CHECK-LABEL: 'test_may_clobber2'
; CHECK-NEXT: loop:
; CHECK-NEXT: Report: unsafe dependent memory operations in loop. Use #pragma clang loop distribute(enable) to allow loop distribution to attempt to isolate the offending operations into a separate loop
; CHECK-NEXT: Backward loop carried data dependence that prevents store-to-load forwarding.
; CHECK-NEXT: Dependences:
; CHECK-NEXT: BackwardVectorizableButPreventsForwarding:
; CHECK-NEXT: %v = load i64, ptr %a1, align 32 ->
; CHECK-NEXT: store i64 %v, ptr %a2, align 32
; CHECK-EMPTY:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv = phi i64 [0, %entry], [%iv.next, %loop]
%a1 = getelementptr i64, ptr %p, i64 %iv
%v = load i64, ptr %a1, align 32
%offset = add i64 %iv, 9
%a2 = getelementptr i64, ptr %p, i64 %offset
store i64 %v, ptr %a2, align 32
%iv.next = add i64 %iv, 1
%cmp = icmp ne i64 %iv, 199
br i1 %cmp, label %loop, label %exit
exit:
ret void
}
define void @test_may_clobber3(ptr %p) {
; CHECK-LABEL: 'test_may_clobber3'
; CHECK-NEXT: loop:
; CHECK-NEXT: Memory dependences are safe with a maximum safe vector width of 640 bits, with a maximum safe store-load forward width of 128 bits
; CHECK-NEXT: Dependences:
; CHECK-NEXT: BackwardVectorizable:
; CHECK-NEXT: %v = load i64, ptr %a1, align 32 ->
; CHECK-NEXT: store i64 %v, ptr %a2, align 32
; CHECK-EMPTY:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv = phi i64 [0, %entry], [%iv.next, %loop]
%a1 = getelementptr i64, ptr %p, i64 %iv
%v = load i64, ptr %a1, align 32
%offset = add i64 %iv, 10
%a2 = getelementptr i64, ptr %p, i64 %offset
store i64 %v, ptr %a2, align 32
%iv.next = add i64 %iv, 1
%cmp = icmp ne i64 %iv, 199
br i1 %cmp, label %loop, label %exit
exit:
ret void
}
; Trivially no overlap due to minimum distance (8192) exceeding value of VLEN and LMUL.
define void @trivial_due_max_vscale(ptr %p) {
; CHECK-LABEL: 'trivial_due_max_vscale'
; CHECK-NEXT: loop:
; CHECK-NEXT: Memory dependences are safe
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv = phi i64 [0, %entry], [%iv.next, %loop]
%a1 = getelementptr i64, ptr %p, i64 %iv
%v = load i64, ptr %a1, align 32
%offset = add i64 %iv, 8192
%a2 = getelementptr i64, ptr %p, i64 %offset
store i64 %v, ptr %a2, align 32
%iv.next = add i64 %iv, 1
%cmp = icmp ne i64 %iv, 199
br i1 %cmp, label %loop, label %exit
exit:
ret void
}
; Dependence distance could be violated via LMUL>=2 or interleaving.
define void @no_high_lmul_or_interleave(ptr %p) {
; CHECK-LABEL: 'no_high_lmul_or_interleave'
; CHECK-NEXT: loop:
; CHECK-NEXT: Memory dependences are safe with a maximum safe vector width of 65536 bits
; CHECK-NEXT: Dependences:
; CHECK-NEXT: BackwardVectorizable:
; CHECK-NEXT: %v = load i64, ptr %a1, align 32 ->
; CHECK-NEXT: store i64 %v, ptr %a2, align 32
; CHECK-EMPTY:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv = phi i64 [0, %entry], [%iv.next, %loop]
%a1 = getelementptr i64, ptr %p, i64 %iv
%v = load i64, ptr %a1, align 32
%offset = add i64 %iv, 1024
%a2 = getelementptr i64, ptr %p, i64 %offset
store i64 %v, ptr %a2, align 32
%iv.next = add i64 %iv, 1
%cmp = icmp ne i64 %iv, 3001
br i1 %cmp, label %loop, label %exit
exit:
ret void
}
define void @non_power_2_storeloadforward(ptr %A) {
; CHECK-LABEL: 'non_power_2_storeloadforward'
; CHECK-NEXT: loop:
; CHECK-NEXT: Report: unsafe dependent memory operations in loop. Use #pragma clang loop distribute(enable) to allow loop distribution to attempt to isolate the offending operations into a separate loop
; CHECK-NEXT: Backward loop carried data dependence that prevents store-to-load forwarding.
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Forward:
; CHECK-NEXT: %3 = load i32, ptr %gep.iv.4, align 4 ->
; CHECK-NEXT: store i32 %add3, ptr %gep.iv, align 4
; CHECK-EMPTY:
; CHECK-NEXT: BackwardVectorizableButPreventsForwarding:
; CHECK-NEXT: %1 = load i32, ptr %gep.iv.sub.3, align 4 ->
; CHECK-NEXT: store i32 %add3, ptr %gep.iv, align 4
; CHECK-EMPTY:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv = phi i64 [ 16, %entry ], [ %iv.next, %loop ]
%0 = add nsw i64 %iv, -3
%gep.iv.sub.3 = getelementptr inbounds i32, ptr %A, i64 %0
%1 = load i32, ptr %gep.iv.sub.3, align 4
%2 = add nsw i64 %iv, 4
%gep.iv.4 = getelementptr inbounds i32, ptr %A, i64 %2
%3 = load i32, ptr %gep.iv.4, align 4
%add3 = add nsw i32 %3, %1
%gep.iv = getelementptr inbounds i32, ptr %A, i64 %iv
store i32 %add3, ptr %gep.iv, align 4
%iv.next = add i64 %iv, 1
%iv.trunc = trunc i64 %iv.next to i32
%exitcond = icmp ne i32 %iv.trunc, 128
br i1 %exitcond, label %loop, label %exit
exit:
ret void
}
|
