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
|
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -passes=loop-distribute -enable-loop-distribute -S < %s | \
; RUN: FileCheck %s --check-prefix=DEFAULT
; RUN: opt -passes=loop-distribute -enable-loop-distribute -S -enable-mem-access-versioning=0 < %s | \
; RUN: FileCheck %s --check-prefix=NO-VERSION
; If we don't speculate stride for 1 we can't distribute along the line
; because we could have a backward dependence:
;
; for (i = 0; i < n; i++) {
; A[i + 1] = A[i] * B[i];
; =======================
; C[i] = D[i] * A[stride * i];
; }
target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-apple-macosx10.10.0"
define void @f(ptr noalias %a,
;
;
; DEFAULT-LABEL: @f(
; DEFAULT-NEXT: entry:
; DEFAULT-NEXT: br label [[FOR_BODY_LVER_CHECK:%.*]]
; DEFAULT: for.body.lver.check:
; DEFAULT-NEXT: [[IDENT_CHECK:%.*]] = icmp ne i64 [[STRIDE:%.*]], 1
; DEFAULT-NEXT: br i1 [[IDENT_CHECK]], label [[FOR_BODY_PH_LVER_ORIG:%.*]], label [[FOR_BODY_PH_LDIST1:%.*]]
; DEFAULT: for.body.ph.lver.orig:
; DEFAULT-NEXT: br label [[FOR_BODY_LVER_ORIG:%.*]]
; DEFAULT: for.body.lver.orig:
; DEFAULT-NEXT: [[IND_LVER_ORIG:%.*]] = phi i64 [ 0, [[FOR_BODY_PH_LVER_ORIG]] ], [ [[ADD_LVER_ORIG:%.*]], [[FOR_BODY_LVER_ORIG]] ]
; DEFAULT-NEXT: [[ARRAYIDXA_LVER_ORIG:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[IND_LVER_ORIG]]
; DEFAULT-NEXT: [[LOADA_LVER_ORIG:%.*]] = load i32, ptr [[ARRAYIDXA_LVER_ORIG]], align 4
; DEFAULT-NEXT: [[ARRAYIDXB_LVER_ORIG:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[IND_LVER_ORIG]]
; DEFAULT-NEXT: [[LOADB_LVER_ORIG:%.*]] = load i32, ptr [[ARRAYIDXB_LVER_ORIG]], align 4
; DEFAULT-NEXT: [[MULA_LVER_ORIG:%.*]] = mul i32 [[LOADB_LVER_ORIG]], [[LOADA_LVER_ORIG]]
; DEFAULT-NEXT: [[ADD_LVER_ORIG]] = add nuw nsw i64 [[IND_LVER_ORIG]], 1
; DEFAULT-NEXT: [[ARRAYIDXA_PLUS_4_LVER_ORIG:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[ADD_LVER_ORIG]]
; DEFAULT-NEXT: store i32 [[MULA_LVER_ORIG]], ptr [[ARRAYIDXA_PLUS_4_LVER_ORIG]], align 4
; DEFAULT-NEXT: [[ARRAYIDXD_LVER_ORIG:%.*]] = getelementptr inbounds i32, ptr [[D:%.*]], i64 [[IND_LVER_ORIG]]
; DEFAULT-NEXT: [[LOADD_LVER_ORIG:%.*]] = load i32, ptr [[ARRAYIDXD_LVER_ORIG]], align 4
; DEFAULT-NEXT: [[MUL_LVER_ORIG:%.*]] = mul i64 [[IND_LVER_ORIG]], [[STRIDE]]
; DEFAULT-NEXT: [[ARRAYIDXSTRIDEDA_LVER_ORIG:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[MUL_LVER_ORIG]]
; DEFAULT-NEXT: [[LOADSTRIDEDA_LVER_ORIG:%.*]] = load i32, ptr [[ARRAYIDXSTRIDEDA_LVER_ORIG]], align 4
; DEFAULT-NEXT: [[MULC_LVER_ORIG:%.*]] = mul i32 [[LOADD_LVER_ORIG]], [[LOADSTRIDEDA_LVER_ORIG]]
; DEFAULT-NEXT: [[ARRAYIDXC_LVER_ORIG:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i64 [[IND_LVER_ORIG]]
; DEFAULT-NEXT: store i32 [[MULC_LVER_ORIG]], ptr [[ARRAYIDXC_LVER_ORIG]], align 4
; DEFAULT-NEXT: [[EXITCOND_LVER_ORIG:%.*]] = icmp eq i64 [[ADD_LVER_ORIG]], 20
; DEFAULT-NEXT: br i1 [[EXITCOND_LVER_ORIG]], label [[FOR_END_LOOPEXIT:%.*]], label [[FOR_BODY_LVER_ORIG]]
; DEFAULT: for.body.ph.ldist1:
; DEFAULT-NEXT: br label [[FOR_BODY_LDIST1:%.*]]
; DEFAULT: for.body.ldist1:
; DEFAULT-NEXT: [[IND_LDIST1:%.*]] = phi i64 [ 0, [[FOR_BODY_PH_LDIST1]] ], [ [[ADD_LDIST1:%.*]], [[FOR_BODY_LDIST1]] ]
; DEFAULT-NEXT: [[ARRAYIDXA_LDIST1:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IND_LDIST1]]
; DEFAULT-NEXT: [[LOADA_LDIST1:%.*]] = load i32, ptr [[ARRAYIDXA_LDIST1]], align 4
; DEFAULT-NEXT: [[ARRAYIDXB_LDIST1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IND_LDIST1]]
; DEFAULT-NEXT: [[LOADB_LDIST1:%.*]] = load i32, ptr [[ARRAYIDXB_LDIST1]], align 4
; DEFAULT-NEXT: [[MULA_LDIST1:%.*]] = mul i32 [[LOADB_LDIST1]], [[LOADA_LDIST1]]
; DEFAULT-NEXT: [[ADD_LDIST1]] = add nuw nsw i64 [[IND_LDIST1]], 1
; DEFAULT-NEXT: [[ARRAYIDXA_PLUS_4_LDIST1:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[ADD_LDIST1]]
; DEFAULT-NEXT: store i32 [[MULA_LDIST1]], ptr [[ARRAYIDXA_PLUS_4_LDIST1]], align 4
; DEFAULT-NEXT: [[EXITCOND_LDIST1:%.*]] = icmp eq i64 [[ADD_LDIST1]], 20
; DEFAULT-NEXT: br i1 [[EXITCOND_LDIST1]], label [[FOR_BODY_PH:%.*]], label [[FOR_BODY_LDIST1]]
; DEFAULT: for.body.ph:
; DEFAULT-NEXT: br label [[FOR_BODY:%.*]]
; DEFAULT: for.body:
; DEFAULT-NEXT: [[IND:%.*]] = phi i64 [ 0, [[FOR_BODY_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
; DEFAULT-NEXT: [[ADD]] = add nuw nsw i64 [[IND]], 1
; DEFAULT-NEXT: [[ARRAYIDXD:%.*]] = getelementptr inbounds i32, ptr [[D]], i64 [[IND]]
; DEFAULT-NEXT: [[LOADD:%.*]] = load i32, ptr [[ARRAYIDXD]], align 4
; DEFAULT-NEXT: [[MUL:%.*]] = mul i64 [[IND]], [[STRIDE]]
; DEFAULT-NEXT: [[ARRAYIDXSTRIDEDA:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[MUL]]
; DEFAULT-NEXT: [[LOADSTRIDEDA:%.*]] = load i32, ptr [[ARRAYIDXSTRIDEDA]], align 4
; DEFAULT-NEXT: [[MULC:%.*]] = mul i32 [[LOADD]], [[LOADSTRIDEDA]]
; DEFAULT-NEXT: [[ARRAYIDXC:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[IND]]
; DEFAULT-NEXT: store i32 [[MULC]], ptr [[ARRAYIDXC]], align 4
; DEFAULT-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[ADD]], 20
; DEFAULT-NEXT: br i1 [[EXITCOND]], label [[FOR_END_LOOPEXIT1:%.*]], label [[FOR_BODY]]
; DEFAULT: for.end.loopexit:
; DEFAULT-NEXT: br label [[FOR_END:%.*]]
; DEFAULT: for.end.loopexit1:
; DEFAULT-NEXT: br label [[FOR_END]]
; DEFAULT: for.end:
; DEFAULT-NEXT: ret void
;
; NO-VERSION-LABEL: @f(
; NO-VERSION-NEXT: entry:
; NO-VERSION-NEXT: br label [[FOR_BODY:%.*]]
; NO-VERSION: for.body:
; NO-VERSION-NEXT: [[IND:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
; NO-VERSION-NEXT: [[ARRAYIDXA:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[IND]]
; NO-VERSION-NEXT: [[LOADA:%.*]] = load i32, ptr [[ARRAYIDXA]], align 4
; NO-VERSION-NEXT: [[ARRAYIDXB:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[IND]]
; NO-VERSION-NEXT: [[LOADB:%.*]] = load i32, ptr [[ARRAYIDXB]], align 4
; NO-VERSION-NEXT: [[MULA:%.*]] = mul i32 [[LOADB]], [[LOADA]]
; NO-VERSION-NEXT: [[ADD]] = add nuw nsw i64 [[IND]], 1
; NO-VERSION-NEXT: [[ARRAYIDXA_PLUS_4:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[ADD]]
; NO-VERSION-NEXT: store i32 [[MULA]], ptr [[ARRAYIDXA_PLUS_4]], align 4
; NO-VERSION-NEXT: [[ARRAYIDXD:%.*]] = getelementptr inbounds i32, ptr [[D:%.*]], i64 [[IND]]
; NO-VERSION-NEXT: [[LOADD:%.*]] = load i32, ptr [[ARRAYIDXD]], align 4
; NO-VERSION-NEXT: [[MUL:%.*]] = mul i64 [[IND]], [[STRIDE:%.*]]
; NO-VERSION-NEXT: [[ARRAYIDXSTRIDEDA:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[MUL]]
; NO-VERSION-NEXT: [[LOADSTRIDEDA:%.*]] = load i32, ptr [[ARRAYIDXSTRIDEDA]], align 4
; NO-VERSION-NEXT: [[MULC:%.*]] = mul i32 [[LOADD]], [[LOADSTRIDEDA]]
; NO-VERSION-NEXT: [[ARRAYIDXC:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i64 [[IND]]
; NO-VERSION-NEXT: store i32 [[MULC]], ptr [[ARRAYIDXC]], align 4
; NO-VERSION-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[ADD]], 20
; NO-VERSION-NEXT: br i1 [[EXITCOND]], label [[FOR_END:%.*]], label [[FOR_BODY]]
; NO-VERSION: for.end:
; NO-VERSION-NEXT: ret void
;
i32* noalias %b,
ptr noalias %c,
ptr noalias %d,
i64 %stride) {
entry:
br label %for.body
for.body: ; preds = %for.body, %entry
%ind = phi i64 [ 0, %entry ], [ %add, %for.body ]
%arrayidxA = getelementptr inbounds i32, ptr %a, i64 %ind
%loadA = load i32, ptr %arrayidxA, align 4
%arrayidxB = getelementptr inbounds i32, ptr %b, i64 %ind
%loadB = load i32, ptr %arrayidxB, align 4
%mulA = mul i32 %loadB, %loadA
%add = add nuw nsw i64 %ind, 1
%arrayidxA_plus_4 = getelementptr inbounds i32, ptr %a, i64 %add
store i32 %mulA, ptr %arrayidxA_plus_4, align 4
%arrayidxD = getelementptr inbounds i32, ptr %d, i64 %ind
%loadD = load i32, ptr %arrayidxD, align 4
%mul = mul i64 %ind, %stride
%arrayidxStridedA = getelementptr inbounds i32, ptr %a, i64 %mul
%loadStridedA = load i32, ptr %arrayidxStridedA, align 4
%mulC = mul i32 %loadD, %loadStridedA
%arrayidxC = getelementptr inbounds i32, ptr %c, i64 %ind
store i32 %mulC, ptr %arrayidxC, align 4
%exitcond = icmp eq i64 %add, 20
br i1 %exitcond, label %for.end, label %for.body
for.end: ; preds = %for.body
ret void
}
|
