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
|
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -passes='default<O3>' -unroll-runtime -S %s | FileCheck %s
target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-apple-macosx"
@b = global [58 x double] zeroinitializer, align 16
@c = global [58 x double] zeroinitializer, align 16
@a = global [58 x double] zeroinitializer, align 16
; Test case for #42332, showing excessive unrolling of vector loop.
define void @test_known_trip_count() {
; CHECK-LABEL: @test_known_trip_count(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr @b, align 16
; CHECK-NEXT: [[WIDE_LOAD3:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 16), align 16
; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <2 x double>, ptr @c, align 16
; CHECK-NEXT: [[WIDE_LOAD5:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 16), align 16
; CHECK-NEXT: [[TMP0:%.*]] = fadd <2 x double> [[WIDE_LOAD]], [[WIDE_LOAD4]]
; CHECK-NEXT: [[TMP1:%.*]] = fadd <2 x double> [[WIDE_LOAD3]], [[WIDE_LOAD5]]
; CHECK-NEXT: store <2 x double> [[TMP0]], ptr @a, align 16
; CHECK-NEXT: store <2 x double> [[TMP1]], ptr getelementptr inbounds (i8, ptr @a, i64 16), align 16
; CHECK-NEXT: [[WIDE_LOAD_1:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 32), align 16
; CHECK-NEXT: [[WIDE_LOAD3_1:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 48), align 16
; CHECK-NEXT: [[WIDE_LOAD4_1:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 32), align 16
; CHECK-NEXT: [[WIDE_LOAD5_1:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 48), align 16
; CHECK-NEXT: [[TMP2:%.*]] = fadd <2 x double> [[WIDE_LOAD_1]], [[WIDE_LOAD4_1]]
; CHECK-NEXT: [[TMP3:%.*]] = fadd <2 x double> [[WIDE_LOAD3_1]], [[WIDE_LOAD5_1]]
; CHECK-NEXT: store <2 x double> [[TMP2]], ptr getelementptr inbounds (i8, ptr @a, i64 32), align 16
; CHECK-NEXT: store <2 x double> [[TMP3]], ptr getelementptr inbounds (i8, ptr @a, i64 48), align 16
; CHECK-NEXT: [[WIDE_LOAD_2:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 64), align 16
; CHECK-NEXT: [[WIDE_LOAD3_2:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 80), align 16
; CHECK-NEXT: [[WIDE_LOAD4_2:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 64), align 16
; CHECK-NEXT: [[WIDE_LOAD5_2:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 80), align 16
; CHECK-NEXT: [[TMP4:%.*]] = fadd <2 x double> [[WIDE_LOAD_2]], [[WIDE_LOAD4_2]]
; CHECK-NEXT: [[TMP5:%.*]] = fadd <2 x double> [[WIDE_LOAD3_2]], [[WIDE_LOAD5_2]]
; CHECK-NEXT: store <2 x double> [[TMP4]], ptr getelementptr inbounds (i8, ptr @a, i64 64), align 16
; CHECK-NEXT: store <2 x double> [[TMP5]], ptr getelementptr inbounds (i8, ptr @a, i64 80), align 16
; CHECK-NEXT: [[WIDE_LOAD_3:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 96), align 16
; CHECK-NEXT: [[WIDE_LOAD3_3:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 112), align 16
; CHECK-NEXT: [[WIDE_LOAD4_3:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 96), align 16
; CHECK-NEXT: [[WIDE_LOAD5_3:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 112), align 16
; CHECK-NEXT: [[TMP6:%.*]] = fadd <2 x double> [[WIDE_LOAD_3]], [[WIDE_LOAD4_3]]
; CHECK-NEXT: [[TMP7:%.*]] = fadd <2 x double> [[WIDE_LOAD3_3]], [[WIDE_LOAD5_3]]
; CHECK-NEXT: store <2 x double> [[TMP6]], ptr getelementptr inbounds (i8, ptr @a, i64 96), align 16
; CHECK-NEXT: store <2 x double> [[TMP7]], ptr getelementptr inbounds (i8, ptr @a, i64 112), align 16
; CHECK-NEXT: [[WIDE_LOAD_4:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 128), align 16
; CHECK-NEXT: [[WIDE_LOAD3_4:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 144), align 16
; CHECK-NEXT: [[WIDE_LOAD4_4:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 128), align 16
; CHECK-NEXT: [[WIDE_LOAD5_4:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 144), align 16
; CHECK-NEXT: [[TMP8:%.*]] = fadd <2 x double> [[WIDE_LOAD_4]], [[WIDE_LOAD4_4]]
; CHECK-NEXT: [[TMP9:%.*]] = fadd <2 x double> [[WIDE_LOAD3_4]], [[WIDE_LOAD5_4]]
; CHECK-NEXT: store <2 x double> [[TMP8]], ptr getelementptr inbounds (i8, ptr @a, i64 128), align 16
; CHECK-NEXT: store <2 x double> [[TMP9]], ptr getelementptr inbounds (i8, ptr @a, i64 144), align 16
; CHECK-NEXT: [[WIDE_LOAD_5:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 160), align 16
; CHECK-NEXT: [[WIDE_LOAD3_5:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 176), align 16
; CHECK-NEXT: [[WIDE_LOAD4_5:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 160), align 16
; CHECK-NEXT: [[WIDE_LOAD5_5:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 176), align 16
; CHECK-NEXT: [[TMP10:%.*]] = fadd <2 x double> [[WIDE_LOAD_5]], [[WIDE_LOAD4_5]]
; CHECK-NEXT: [[TMP11:%.*]] = fadd <2 x double> [[WIDE_LOAD3_5]], [[WIDE_LOAD5_5]]
; CHECK-NEXT: store <2 x double> [[TMP10]], ptr getelementptr inbounds (i8, ptr @a, i64 160), align 16
; CHECK-NEXT: store <2 x double> [[TMP11]], ptr getelementptr inbounds (i8, ptr @a, i64 176), align 16
; CHECK-NEXT: [[WIDE_LOAD_6:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 192), align 16
; CHECK-NEXT: [[WIDE_LOAD3_6:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 208), align 16
; CHECK-NEXT: [[WIDE_LOAD4_6:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 192), align 16
; CHECK-NEXT: [[WIDE_LOAD5_6:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 208), align 16
; CHECK-NEXT: [[TMP12:%.*]] = fadd <2 x double> [[WIDE_LOAD_6]], [[WIDE_LOAD4_6]]
; CHECK-NEXT: [[TMP13:%.*]] = fadd <2 x double> [[WIDE_LOAD3_6]], [[WIDE_LOAD5_6]]
; CHECK-NEXT: store <2 x double> [[TMP12]], ptr getelementptr inbounds (i8, ptr @a, i64 192), align 16
; CHECK-NEXT: store <2 x double> [[TMP13]], ptr getelementptr inbounds (i8, ptr @a, i64 208), align 16
; CHECK-NEXT: [[WIDE_LOAD_7:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 224), align 16
; CHECK-NEXT: [[WIDE_LOAD3_7:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 240), align 16
; CHECK-NEXT: [[WIDE_LOAD4_7:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 224), align 16
; CHECK-NEXT: [[WIDE_LOAD5_7:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 240), align 16
; CHECK-NEXT: [[TMP14:%.*]] = fadd <2 x double> [[WIDE_LOAD_7]], [[WIDE_LOAD4_7]]
; CHECK-NEXT: [[TMP15:%.*]] = fadd <2 x double> [[WIDE_LOAD3_7]], [[WIDE_LOAD5_7]]
; CHECK-NEXT: store <2 x double> [[TMP14]], ptr getelementptr inbounds (i8, ptr @a, i64 224), align 16
; CHECK-NEXT: store <2 x double> [[TMP15]], ptr getelementptr inbounds (i8, ptr @a, i64 240), align 16
; CHECK-NEXT: [[WIDE_LOAD_8:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 256), align 16
; CHECK-NEXT: [[WIDE_LOAD3_8:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 272), align 16
; CHECK-NEXT: [[WIDE_LOAD4_8:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 256), align 16
; CHECK-NEXT: [[WIDE_LOAD5_8:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 272), align 16
; CHECK-NEXT: [[TMP16:%.*]] = fadd <2 x double> [[WIDE_LOAD_8]], [[WIDE_LOAD4_8]]
; CHECK-NEXT: [[TMP17:%.*]] = fadd <2 x double> [[WIDE_LOAD3_8]], [[WIDE_LOAD5_8]]
; CHECK-NEXT: store <2 x double> [[TMP16]], ptr getelementptr inbounds (i8, ptr @a, i64 256), align 16
; CHECK-NEXT: store <2 x double> [[TMP17]], ptr getelementptr inbounds (i8, ptr @a, i64 272), align 16
; CHECK-NEXT: [[WIDE_LOAD_9:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 288), align 16
; CHECK-NEXT: [[WIDE_LOAD3_9:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 304), align 16
; CHECK-NEXT: [[WIDE_LOAD4_9:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 288), align 16
; CHECK-NEXT: [[WIDE_LOAD5_9:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 304), align 16
; CHECK-NEXT: [[TMP18:%.*]] = fadd <2 x double> [[WIDE_LOAD_9]], [[WIDE_LOAD4_9]]
; CHECK-NEXT: [[TMP19:%.*]] = fadd <2 x double> [[WIDE_LOAD3_9]], [[WIDE_LOAD5_9]]
; CHECK-NEXT: store <2 x double> [[TMP18]], ptr getelementptr inbounds (i8, ptr @a, i64 288), align 16
; CHECK-NEXT: store <2 x double> [[TMP19]], ptr getelementptr inbounds (i8, ptr @a, i64 304), align 16
; CHECK-NEXT: [[WIDE_LOAD_10:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 320), align 16
; CHECK-NEXT: [[WIDE_LOAD3_10:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 336), align 16
; CHECK-NEXT: [[WIDE_LOAD4_10:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 320), align 16
; CHECK-NEXT: [[WIDE_LOAD5_10:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 336), align 16
; CHECK-NEXT: [[TMP20:%.*]] = fadd <2 x double> [[WIDE_LOAD_10]], [[WIDE_LOAD4_10]]
; CHECK-NEXT: [[TMP21:%.*]] = fadd <2 x double> [[WIDE_LOAD3_10]], [[WIDE_LOAD5_10]]
; CHECK-NEXT: store <2 x double> [[TMP20]], ptr getelementptr inbounds (i8, ptr @a, i64 320), align 16
; CHECK-NEXT: store <2 x double> [[TMP21]], ptr getelementptr inbounds (i8, ptr @a, i64 336), align 16
; CHECK-NEXT: [[WIDE_LOAD_11:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 352), align 16
; CHECK-NEXT: [[WIDE_LOAD3_11:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 368), align 16
; CHECK-NEXT: [[WIDE_LOAD4_11:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 352), align 16
; CHECK-NEXT: [[WIDE_LOAD5_11:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 368), align 16
; CHECK-NEXT: [[TMP22:%.*]] = fadd <2 x double> [[WIDE_LOAD_11]], [[WIDE_LOAD4_11]]
; CHECK-NEXT: [[TMP23:%.*]] = fadd <2 x double> [[WIDE_LOAD3_11]], [[WIDE_LOAD5_11]]
; CHECK-NEXT: store <2 x double> [[TMP22]], ptr getelementptr inbounds (i8, ptr @a, i64 352), align 16
; CHECK-NEXT: store <2 x double> [[TMP23]], ptr getelementptr inbounds (i8, ptr @a, i64 368), align 16
; CHECK-NEXT: [[WIDE_LOAD_12:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 384), align 16
; CHECK-NEXT: [[WIDE_LOAD3_12:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 400), align 16
; CHECK-NEXT: [[WIDE_LOAD4_12:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 384), align 16
; CHECK-NEXT: [[WIDE_LOAD5_12:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 400), align 16
; CHECK-NEXT: [[TMP24:%.*]] = fadd <2 x double> [[WIDE_LOAD_12]], [[WIDE_LOAD4_12]]
; CHECK-NEXT: [[TMP25:%.*]] = fadd <2 x double> [[WIDE_LOAD3_12]], [[WIDE_LOAD5_12]]
; CHECK-NEXT: store <2 x double> [[TMP24]], ptr getelementptr inbounds (i8, ptr @a, i64 384), align 16
; CHECK-NEXT: store <2 x double> [[TMP25]], ptr getelementptr inbounds (i8, ptr @a, i64 400), align 16
; CHECK-NEXT: [[WIDE_LOAD_13:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 416), align 16
; CHECK-NEXT: [[WIDE_LOAD3_13:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 432), align 16
; CHECK-NEXT: [[WIDE_LOAD4_13:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 416), align 16
; CHECK-NEXT: [[WIDE_LOAD5_13:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 432), align 16
; CHECK-NEXT: [[TMP26:%.*]] = fadd <2 x double> [[WIDE_LOAD_13]], [[WIDE_LOAD4_13]]
; CHECK-NEXT: [[TMP27:%.*]] = fadd <2 x double> [[WIDE_LOAD3_13]], [[WIDE_LOAD5_13]]
; CHECK-NEXT: store <2 x double> [[TMP26]], ptr getelementptr inbounds (i8, ptr @a, i64 416), align 16
; CHECK-NEXT: store <2 x double> [[TMP27]], ptr getelementptr inbounds (i8, ptr @a, i64 432), align 16
; CHECK-NEXT: [[WIDE_LOAD_14:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 448), align 16
; CHECK-NEXT: [[WIDE_LOAD3_14:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @b, i64 464), align 16
; CHECK-NEXT: [[WIDE_LOAD4_14:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 448), align 16
; CHECK-NEXT: [[WIDE_LOAD5_14:%.*]] = load <2 x double>, ptr getelementptr inbounds (i8, ptr @c, i64 464), align 16
; CHECK-NEXT: [[TMP28:%.*]] = fadd <2 x double> [[WIDE_LOAD_14]], [[WIDE_LOAD4_14]]
; CHECK-NEXT: [[TMP29:%.*]] = fadd <2 x double> [[WIDE_LOAD3_14]], [[WIDE_LOAD5_14]]
; CHECK-NEXT: store <2 x double> [[TMP28]], ptr getelementptr inbounds (i8, ptr @a, i64 448), align 16
; CHECK-NEXT: store <2 x double> [[TMP29]], ptr getelementptr inbounds (i8, ptr @a, i64 464), align 16
; CHECK-NEXT: [[TMP30:%.*]] = load double, ptr getelementptr inbounds (i8, ptr @b, i64 480), align 16
; CHECK-NEXT: [[TMP31:%.*]] = load double, ptr getelementptr inbounds (i8, ptr @c, i64 480), align 16
; CHECK-NEXT: [[ADD:%.*]] = fadd double [[TMP30]], [[TMP31]]
; CHECK-NEXT: store double [[ADD]], ptr getelementptr inbounds (i8, ptr @a, i64 480), align 16
; CHECK-NEXT: ret void
;
entry:
br label %for.cond
for.cond:
%i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
%cmp = icmp slt i32 %i.0, 61
br i1 %cmp, label %for.body, label %exit
for.body:
%idxprom = sext i32 %i.0 to i64
%arrayidx = getelementptr inbounds [58 x double], ptr @b, i64 0, i64 %idxprom
%0 = load double, ptr %arrayidx, align 8
%idxprom1 = sext i32 %i.0 to i64
%arrayidx2 = getelementptr inbounds [58 x double], ptr @c, i64 0, i64 %idxprom1
%1 = load double, ptr %arrayidx2, align 8
%add = fadd double %0, %1
%idxprom3 = sext i32 %i.0 to i64
%arrayidx4 = getelementptr inbounds [58 x double], ptr @a, i64 0, i64 %idxprom3
store double %add, ptr %arrayidx4, align 8
%inc = add nsw i32 %i.0, 1
br label %for.cond
exit:
ret void
}
define void @test_runtime_trip_count(i32 %N) {
; CHECK-LABEL: @test_runtime_trip_count(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[CMP1:%.*]] = icmp sgt i32 [[N:%.*]], 0
; CHECK-NEXT: br i1 [[CMP1]], label [[FOR_BODY_PREHEADER:%.*]], label [[EXIT:%.*]]
; CHECK: for.body.preheader:
; CHECK-NEXT: [[WIDE_TRIP_COUNT:%.*]] = zext nneg i32 [[N]] to i64
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[N]], 4
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[FOR_BODY_PREHEADER7:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[WIDE_TRIP_COUNT]], 2147483644
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds [58 x double], ptr @b, i64 0, i64 [[INDEX]]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP0]], align 16
; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <2 x double>, ptr [[TMP1]], align 16
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [58 x double], ptr @c, i64 0, i64 [[INDEX]]
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 16
; CHECK-NEXT: [[WIDE_LOAD5:%.*]] = load <2 x double>, ptr [[TMP2]], align 16
; CHECK-NEXT: [[WIDE_LOAD6:%.*]] = load <2 x double>, ptr [[TMP3]], align 16
; CHECK-NEXT: [[TMP4:%.*]] = fadd <2 x double> [[WIDE_LOAD]], [[WIDE_LOAD5]]
; CHECK-NEXT: [[TMP5:%.*]] = fadd <2 x double> [[WIDE_LOAD4]], [[WIDE_LOAD6]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [58 x double], ptr @a, i64 0, i64 [[INDEX]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP6]], i64 16
; CHECK-NEXT: store <2 x double> [[TMP4]], ptr [[TMP6]], align 16
; CHECK-NEXT: store <2 x double> [[TMP5]], ptr [[TMP7]], align 16
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[WIDE_TRIP_COUNT]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[EXIT]], label [[FOR_BODY_PREHEADER7]]
; CHECK: for.body.preheader7:
; CHECK-NEXT: [[INDVARS_IV_PH:%.*]] = phi i64 [ 0, [[FOR_BODY_PREHEADER]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[INDVARS_IV_PH]], [[FOR_BODY_PREHEADER7]] ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [58 x double], ptr @b, i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT: [[TMP9:%.*]] = load double, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds [58 x double], ptr @c, i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT: [[TMP10:%.*]] = load double, ptr [[ARRAYIDX2]], align 8
; CHECK-NEXT: [[ADD:%.*]] = fadd double [[TMP9]], [[TMP10]]
; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [58 x double], ptr @a, i64 0, i64 [[INDVARS_IV]]
; CHECK-NEXT: store double [[ADD]], ptr [[ARRAYIDX4]], align 8
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[WIDE_TRIP_COUNT]]
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[EXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: exit:
; CHECK-NEXT: ret void
;
entry:
br label %for.cond
for.cond:
%i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
%cmp = icmp slt i32 %i.0, %N
br i1 %cmp, label %for.body, label %exit
for.body:
%idxprom = sext i32 %i.0 to i64
%arrayidx = getelementptr inbounds [58 x double], ptr @b, i64 0, i64 %idxprom
%0 = load double, ptr %arrayidx, align 8
%idxprom1 = sext i32 %i.0 to i64
%arrayidx2 = getelementptr inbounds [58 x double], ptr @c, i64 0, i64 %idxprom1
%1 = load double, ptr %arrayidx2, align 8
%add = fadd double %0, %1
%idxprom3 = sext i32 %i.0 to i64
%arrayidx4 = getelementptr inbounds [58 x double], ptr @a, i64 0, i64 %idxprom3
store double %add, ptr %arrayidx4, align 8
%inc = add nsw i32 %i.0, 1
br label %for.cond
exit:
ret void
}
|
