matmul.polly.interchanged.ll
13.2 KB
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
; ModuleID = '<stdin>'
source_filename = "matmul.c"
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
%struct._IO_FILE = type { i32, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, %struct._IO_marker*, %struct._IO_FILE*, i32, i32, i64, i16, i8, [1 x i8], i8*, i64, i8*, i8*, i8*, i8*, i64, i32, [20 x i8] }
%struct._IO_marker = type { %struct._IO_marker*, %struct._IO_FILE*, i32 }
@A = common dso_local local_unnamed_addr global [1536 x [1536 x float]] zeroinitializer, align 16
@B = common dso_local local_unnamed_addr global [1536 x [1536 x float]] zeroinitializer, align 16
@stdout = external dso_local local_unnamed_addr global %struct._IO_FILE*, align 8
@.str = private unnamed_addr constant [5 x i8] c"%lf \00", align 1
@C = common dso_local local_unnamed_addr global [1536 x [1536 x float]] zeroinitializer, align 16
; Function Attrs: noinline norecurse nounwind uwtable writeonly
define dso_local void @init_array() local_unnamed_addr #0 {
entry:
br label %polly.loop_header
polly.exiting: ; preds = %polly.loop_exit3
ret void
polly.loop_header: ; preds = %polly.loop_exit3, %entry
%polly.indvar = phi i64 [ 0, %entry ], [ %polly.indvar_next, %polly.loop_exit3 ]
%0 = trunc i64 %polly.indvar to i32
br label %polly.loop_header1
polly.loop_exit3: ; preds = %polly.loop_header1
%polly.indvar_next = add nuw nsw i64 %polly.indvar, 1
%exitcond1 = icmp eq i64 %polly.indvar_next, 1536
br i1 %exitcond1, label %polly.exiting, label %polly.loop_header
polly.loop_header1: ; preds = %polly.loop_header1, %polly.loop_header
%polly.indvar4 = phi i64 [ 0, %polly.loop_header ], [ %polly.indvar_next5.1, %polly.loop_header1 ]
%1 = trunc i64 %polly.indvar4 to i32
%2 = mul nuw nsw i32 %1, %0
%3 = and i32 %2, 1022
%4 = or i32 %3, 1
%p_conv = sitofp i32 %4 to double
%p_div = fmul double %p_conv, 5.000000e-01
%p_conv4 = fptrunc double %p_div to float
%scevgep7 = getelementptr [1536 x [1536 x float]], [1536 x [1536 x float]]* @A, i64 0, i64 %polly.indvar, i64 %polly.indvar4
store float %p_conv4, float* %scevgep7, align 8, !alias.scope !2, !noalias !4
%scevgep9 = getelementptr [1536 x [1536 x float]], [1536 x [1536 x float]]* @B, i64 0, i64 %polly.indvar, i64 %polly.indvar4
store float %p_conv4, float* %scevgep9, align 8, !alias.scope !5, !noalias !6
%polly.indvar_next5 = or i64 %polly.indvar4, 1
%5 = trunc i64 %polly.indvar_next5 to i32
%6 = mul nuw nsw i32 %5, %0
%7 = and i32 %6, 1023
%8 = add nuw nsw i32 %7, 1
%p_conv.1 = sitofp i32 %8 to double
%p_div.1 = fmul double %p_conv.1, 5.000000e-01
%p_conv4.1 = fptrunc double %p_div.1 to float
%scevgep7.1 = getelementptr [1536 x [1536 x float]], [1536 x [1536 x float]]* @A, i64 0, i64 %polly.indvar, i64 %polly.indvar_next5
store float %p_conv4.1, float* %scevgep7.1, align 4, !alias.scope !2, !noalias !4
%scevgep9.1 = getelementptr [1536 x [1536 x float]], [1536 x [1536 x float]]* @B, i64 0, i64 %polly.indvar, i64 %polly.indvar_next5
store float %p_conv4.1, float* %scevgep9.1, align 4, !alias.scope !5, !noalias !6
%polly.indvar_next5.1 = add nuw nsw i64 %polly.indvar4, 2
%exitcond.1 = icmp eq i64 %polly.indvar_next5.1, 1536
br i1 %exitcond.1, label %polly.loop_exit3, label %polly.loop_header1
}
; Function Attrs: noinline nounwind uwtable
define dso_local void @print_array() local_unnamed_addr #1 {
entry:
br label %for.cond1.preheader
for.cond1.preheader: ; preds = %for.end, %entry
%indvars.iv6 = phi i64 [ 0, %entry ], [ %indvars.iv.next7, %for.end ]
%0 = load %struct._IO_FILE*, %struct._IO_FILE** @stdout, align 8
br label %for.body3
for.body3: ; preds = %for.inc, %for.cond1.preheader
%indvars.iv = phi i64 [ 0, %for.cond1.preheader ], [ %indvars.iv.next, %for.inc ]
%1 = phi %struct._IO_FILE* [ %0, %for.cond1.preheader ], [ %5, %for.inc ]
%arrayidx5 = getelementptr inbounds [1536 x [1536 x float]], [1536 x [1536 x float]]* @C, i64 0, i64 %indvars.iv6, i64 %indvars.iv
%2 = load float, float* %arrayidx5, align 4
%conv = fpext float %2 to double
%call = tail call i32 (%struct._IO_FILE*, i8*, ...) @fprintf(%struct._IO_FILE* %1, i8* getelementptr inbounds ([5 x i8], [5 x i8]* @.str, i64 0, i64 0), double %conv) #4
%3 = trunc i64 %indvars.iv to i32
%rem = urem i32 %3, 80
%cmp6 = icmp eq i32 %rem, 79
br i1 %cmp6, label %if.then, label %for.inc
if.then: ; preds = %for.body3
%4 = load %struct._IO_FILE*, %struct._IO_FILE** @stdout, align 8
%fputc3 = tail call i32 @fputc(i32 10, %struct._IO_FILE* %4)
br label %for.inc
for.inc: ; preds = %if.then, %for.body3
%indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
%5 = load %struct._IO_FILE*, %struct._IO_FILE** @stdout, align 8
%exitcond = icmp eq i64 %indvars.iv.next, 1536
br i1 %exitcond, label %for.end, label %for.body3
for.end: ; preds = %for.inc
%fputc = tail call i32 @fputc(i32 10, %struct._IO_FILE* %5)
%indvars.iv.next7 = add nuw nsw i64 %indvars.iv6, 1
%exitcond8 = icmp eq i64 %indvars.iv.next7, 1536
br i1 %exitcond8, label %for.end12, label %for.cond1.preheader
for.end12: ; preds = %for.end
ret void
}
; Function Attrs: nounwind
declare dso_local i32 @fprintf(%struct._IO_FILE* nocapture, i8* nocapture readonly, ...) local_unnamed_addr #2
; Function Attrs: noinline norecurse nounwind uwtable
define dso_local i32 @main() local_unnamed_addr #3 {
entry:
tail call void @init_array()
call void @llvm.memset.p0i8.i64(i8* align 16 bitcast ([1536 x [1536 x float]]* @C to i8*), i8 0, i64 9437184, i1 false)
br label %polly.loop_header8
polly.exiting: ; preds = %polly.loop_exit16
ret i32 0
polly.loop_header8: ; preds = %polly.loop_exit16, %entry
%polly.indvar11 = phi i64 [ %polly.indvar_next12, %polly.loop_exit16 ], [ 0, %entry ]
br label %polly.loop_header14
polly.loop_exit16: ; preds = %polly.loop_exit22
%polly.indvar_next12 = add nuw nsw i64 %polly.indvar11, 1
%exitcond2 = icmp eq i64 %polly.indvar_next12, 1536
br i1 %exitcond2, label %polly.exiting, label %polly.loop_header8
polly.loop_header14: ; preds = %polly.loop_exit22, %polly.loop_header8
%polly.indvar17 = phi i64 [ 0, %polly.loop_header8 ], [ %polly.indvar_next18, %polly.loop_exit22 ]
%scevgep29 = getelementptr [1536 x [1536 x float]], [1536 x [1536 x float]]* @A, i64 0, i64 %polly.indvar11, i64 %polly.indvar17
%_p_scalar_30 = load float, float* %scevgep29, align 4, !alias.scope !7, !noalias !9
%broadcast.splatinsert10 = insertelement <4 x float> undef, float %_p_scalar_30, i32 0
%broadcast.splat11 = shufflevector <4 x float> %broadcast.splatinsert10, <4 x float> undef, <4 x i32> zeroinitializer
%broadcast.splatinsert12 = insertelement <4 x float> undef, float %_p_scalar_30, i32 0
%broadcast.splat13 = shufflevector <4 x float> %broadcast.splatinsert12, <4 x float> undef, <4 x i32> zeroinitializer
br label %vector.body
vector.body: ; preds = %vector.body, %polly.loop_header14
%index = phi i64 [ 0, %polly.loop_header14 ], [ %index.next.1, %vector.body ]
%0 = getelementptr [1536 x [1536 x float]], [1536 x [1536 x float]]* @C, i64 0, i64 %polly.indvar11, i64 %index
%1 = bitcast float* %0 to <4 x float>*
%wide.load = load <4 x float>, <4 x float>* %1, align 16, !alias.scope !10, !noalias !12
%2 = getelementptr float, float* %0, i64 4
%3 = bitcast float* %2 to <4 x float>*
%wide.load7 = load <4 x float>, <4 x float>* %3, align 16, !alias.scope !10, !noalias !12
%4 = getelementptr [1536 x [1536 x float]], [1536 x [1536 x float]]* @B, i64 0, i64 %polly.indvar17, i64 %index
%5 = bitcast float* %4 to <4 x float>*
%wide.load8 = load <4 x float>, <4 x float>* %5, align 16, !alias.scope !11, !noalias !13
%6 = getelementptr float, float* %4, i64 4
%7 = bitcast float* %6 to <4 x float>*
%wide.load9 = load <4 x float>, <4 x float>* %7, align 16, !alias.scope !11, !noalias !13
%8 = fmul <4 x float> %broadcast.splat11, %wide.load8
%9 = fmul <4 x float> %broadcast.splat13, %wide.load9
%10 = fadd <4 x float> %wide.load, %8
%11 = fadd <4 x float> %wide.load7, %9
%12 = bitcast float* %0 to <4 x float>*
store <4 x float> %10, <4 x float>* %12, align 16, !alias.scope !10, !noalias !12
%13 = bitcast float* %2 to <4 x float>*
store <4 x float> %11, <4 x float>* %13, align 16, !alias.scope !10, !noalias !12
%index.next = or i64 %index, 8
%14 = getelementptr [1536 x [1536 x float]], [1536 x [1536 x float]]* @C, i64 0, i64 %polly.indvar11, i64 %index.next
%15 = bitcast float* %14 to <4 x float>*
%wide.load.1 = load <4 x float>, <4 x float>* %15, align 16, !alias.scope !10, !noalias !12
%16 = getelementptr float, float* %14, i64 4
%17 = bitcast float* %16 to <4 x float>*
%wide.load7.1 = load <4 x float>, <4 x float>* %17, align 16, !alias.scope !10, !noalias !12
%18 = getelementptr [1536 x [1536 x float]], [1536 x [1536 x float]]* @B, i64 0, i64 %polly.indvar17, i64 %index.next
%19 = bitcast float* %18 to <4 x float>*
%wide.load8.1 = load <4 x float>, <4 x float>* %19, align 16, !alias.scope !11, !noalias !13
%20 = getelementptr float, float* %18, i64 4
%21 = bitcast float* %20 to <4 x float>*
%wide.load9.1 = load <4 x float>, <4 x float>* %21, align 16, !alias.scope !11, !noalias !13
%22 = fmul <4 x float> %broadcast.splat11, %wide.load8.1
%23 = fmul <4 x float> %broadcast.splat13, %wide.load9.1
%24 = fadd <4 x float> %wide.load.1, %22
%25 = fadd <4 x float> %wide.load7.1, %23
%26 = bitcast float* %14 to <4 x float>*
store <4 x float> %24, <4 x float>* %26, align 16, !alias.scope !10, !noalias !12
%27 = bitcast float* %16 to <4 x float>*
store <4 x float> %25, <4 x float>* %27, align 16, !alias.scope !10, !noalias !12
%index.next.1 = add nuw nsw i64 %index, 16
%28 = icmp eq i64 %index.next.1, 1536
br i1 %28, label %polly.loop_exit22, label %vector.body, !llvm.loop !14
polly.loop_exit22: ; preds = %vector.body
%polly.indvar_next18 = add nuw nsw i64 %polly.indvar17, 1
%exitcond1 = icmp eq i64 %polly.indvar_next18, 1536
br i1 %exitcond1, label %polly.loop_exit16, label %polly.loop_header14
}
; Function Attrs: nounwind
declare i32 @fputc(i32, %struct._IO_FILE* nocapture) local_unnamed_addr #4
; Function Attrs: argmemonly nounwind
declare void @llvm.memset.p0i8.i64(i8* nocapture writeonly, i8, i64, i1) #5
attributes #0 = { noinline norecurse nounwind uwtable writeonly "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "polly-optimized" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+fxsr,+mmx,+sse,+sse2,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" }
attributes #1 = { noinline nounwind uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+fxsr,+mmx,+sse,+sse2,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" }
attributes #2 = { nounwind "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+fxsr,+mmx,+sse,+sse2,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" }
attributes #3 = { noinline norecurse nounwind uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "polly-optimized" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+fxsr,+mmx,+sse,+sse2,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" }
attributes #4 = { nounwind }
attributes #5 = { argmemonly nounwind }
!llvm.module.flags = !{!0}
!llvm.ident = !{!1}
!0 = !{i32 1, !"wchar_size", i32 4}
!1 = !{!"clang version 8.0.0 (trunk 342834) (llvm/trunk 342856)"}
!2 = distinct !{!2, !3, !"polly.alias.scope.MemRef_A"}
!3 = distinct !{!3, !"polly.alias.scope.domain"}
!4 = !{!5}
!5 = distinct !{!5, !3, !"polly.alias.scope.MemRef_B"}
!6 = !{!2}
!7 = distinct !{!7, !8, !"polly.alias.scope.MemRef_A"}
!8 = distinct !{!8, !"polly.alias.scope.domain"}
!9 = !{!10, !11}
!10 = distinct !{!10, !8, !"polly.alias.scope.MemRef_C"}
!11 = distinct !{!11, !8, !"polly.alias.scope.MemRef_B"}
!12 = !{!7, !11}
!13 = !{!10, !7}
!14 = distinct !{!14, !15}
!15 = !{!"llvm.loop.isvectorized", i32 1}