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
|
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -passes=loop-vectorize < %s -S -o - | FileCheck %s
target datalayout = "e-m:e-p:32:32-Fi8-i64:64-v128:64:128-a:0:32-n32-S64"
target triple = "thumbv8.1m.main-none-eabi"
; Test cases to make sure LV & loop versioning can handle loops with
; multiple exiting branches.
; Multiple branches exiting the loop to a unique exit block.
define void @multiple_exits_unique_exit_block(ptr %A, ptr %B, i32 %N) #0 {
; CHECK-LABEL: @multiple_exits_unique_exit_block(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[A2:%.*]] = ptrtoint ptr [[A:%.*]] to i32
; CHECK-NEXT: [[B1:%.*]] = ptrtoint ptr [[B:%.*]] to i32
; CHECK-NEXT: [[UMIN:%.*]] = call i32 @llvm.umin.i32(i32 [[N:%.*]], i32 999)
; CHECK-NEXT: [[TMP0:%.*]] = add nuw nsw i32 [[UMIN]], 1
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ule i32 [[TMP0]], 4
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_MEMCHECK:%.*]]
; CHECK: vector.memcheck:
; CHECK-NEXT: [[TMP1:%.*]] = sub i32 [[B1]], [[A2]]
; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i32 [[TMP1]], 16
; CHECK-NEXT: br i1 [[DIFF_CHECK]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i32 [[TMP0]], 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i32 [[N_MOD_VF]], 0
; CHECK-NEXT: [[TMP3:%.*]] = select i1 [[TMP2]], i32 4, i32 [[N_MOD_VF]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i32 [[TMP0]], [[TMP3]]
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP4:%.*]] = add i32 [[INDEX]], 0
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP4]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP5]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP6]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP4]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[TMP7]], i32 0
; CHECK-NEXT: store <4 x i32> [[WIDE_LOAD]], ptr [[TMP8]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ], [ 0, [[VECTOR_MEMCHECK]] ]
; CHECK-NEXT: br label [[LOOP_HEADER:%.*]]
; CHECK: loop.header:
; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY:%.*]] ]
; CHECK-NEXT: [[COND_0:%.*]] = icmp eq i32 [[IV]], [[N]]
; CHECK-NEXT: br i1 [[COND_0]], label [[EXIT:%.*]], label [[FOR_BODY]]
; CHECK: for.body:
; CHECK-NEXT: [[A_GEP:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[IV]]
; CHECK-NEXT: [[LV:%.*]] = load i32, ptr [[A_GEP]], align 4
; CHECK-NEXT: [[B_GEP:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[IV]]
; CHECK-NEXT: store i32 [[LV]], ptr [[B_GEP]], align 4
; CHECK-NEXT: [[IV_NEXT]] = add nuw i32 [[IV]], 1
; CHECK-NEXT: [[COND_1:%.*]] = icmp ult i32 [[IV_NEXT]], 1000
; CHECK-NEXT: br i1 [[COND_1]], label [[LOOP_HEADER]], label [[EXIT]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: exit:
; CHECK-NEXT: ret void
;
entry:
br label %loop.header
loop.header:
%iv = phi i32 [ 0, %entry ], [ %iv.next, %for.body ]
%cond.0 = icmp eq i32 %iv, %N
br i1 %cond.0, label %exit, label %for.body
for.body:
%A.gep = getelementptr inbounds i32, ptr %A, i32 %iv
%lv = load i32, ptr %A.gep, align 4
%B.gep = getelementptr inbounds i32, ptr %B, i32 %iv
store i32 %lv, ptr %B.gep, align 4
%iv.next = add nuw i32 %iv, 1
%cond.1 = icmp ult i32 %iv.next, 1000
br i1 %cond.1, label %loop.header, label %exit
exit:
ret void
}
; Multiple branches exiting the loop to different blocks.
define i32 @multiple_exits_multiple_exit_blocks(ptr %A, ptr %B, i32 %N) #0 {
; CHECK-LABEL: @multiple_exits_multiple_exit_blocks(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[A2:%.*]] = ptrtoint ptr [[A:%.*]] to i32
; CHECK-NEXT: [[B1:%.*]] = ptrtoint ptr [[B:%.*]] to i32
; CHECK-NEXT: [[UMIN:%.*]] = call i32 @llvm.umin.i32(i32 [[N:%.*]], i32 999)
; CHECK-NEXT: [[TMP0:%.*]] = add nuw nsw i32 [[UMIN]], 1
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ule i32 [[TMP0]], 4
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_MEMCHECK:%.*]]
; CHECK: vector.memcheck:
; CHECK-NEXT: [[TMP1:%.*]] = sub i32 [[B1]], [[A2]]
; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i32 [[TMP1]], 16
; CHECK-NEXT: br i1 [[DIFF_CHECK]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i32 [[TMP0]], 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i32 [[N_MOD_VF]], 0
; CHECK-NEXT: [[TMP3:%.*]] = select i1 [[TMP2]], i32 4, i32 [[N_MOD_VF]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i32 [[TMP0]], [[TMP3]]
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP4:%.*]] = add i32 [[INDEX]], 0
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP4]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP5]], i32 0
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP6]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP4]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[TMP7]], i32 0
; CHECK-NEXT: store <4 x i32> [[WIDE_LOAD]], ptr [[TMP8]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ], [ 0, [[VECTOR_MEMCHECK]] ]
; CHECK-NEXT: br label [[LOOP_HEADER:%.*]]
; CHECK: loop.header:
; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY:%.*]] ]
; CHECK-NEXT: [[COND_0:%.*]] = icmp eq i32 [[IV]], [[N]]
; CHECK-NEXT: br i1 [[COND_0]], label [[EXIT_0:%.*]], label [[FOR_BODY]]
; CHECK: for.body:
; CHECK-NEXT: [[A_GEP:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[IV]]
; CHECK-NEXT: [[LV:%.*]] = load i32, ptr [[A_GEP]], align 4
; CHECK-NEXT: [[B_GEP:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[IV]]
; CHECK-NEXT: store i32 [[LV]], ptr [[B_GEP]], align 4
; CHECK-NEXT: [[IV_NEXT]] = add nuw i32 [[IV]], 1
; CHECK-NEXT: [[COND_1:%.*]] = icmp ult i32 [[IV_NEXT]], 1000
; CHECK-NEXT: br i1 [[COND_1]], label [[LOOP_HEADER]], label [[EXIT_1:%.*]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK: exit.0:
; CHECK-NEXT: ret i32 1
; CHECK: exit.1:
; CHECK-NEXT: ret i32 2
;
entry:
br label %loop.header
loop.header:
%iv = phi i32 [ 0, %entry ], [ %iv.next, %for.body ]
%cond.0 = icmp eq i32 %iv, %N
br i1 %cond.0, label %exit.0, label %for.body
for.body:
%A.gep = getelementptr inbounds i32, ptr %A, i32 %iv
%lv = load i32, ptr %A.gep, align 4
%B.gep = getelementptr inbounds i32, ptr %B, i32 %iv
store i32 %lv, ptr %B.gep, align 4
%iv.next = add nuw i32 %iv, 1
%cond.1 = icmp ult i32 %iv.next, 1000
br i1 %cond.1, label %loop.header, label %exit.1
exit.0:
ret i32 1
exit.1:
ret i32 2
}
attributes #0 = { "target-features"="+mve" }
|
