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
|
; REQUIRES: asserts
; RUN: opt -mtriple=s390x-unknown-linux -mcpu=z13 -passes=loop-vectorize \
; RUN: -force-vector-width=4 -debug-only=loop-vectorize,vectorutils \
; RUN: -disable-output < %s 2>&1 | FileCheck %s
;
; Check that the loop vectorizer performs memory interleaving with accurate
; cost estimations.
; Simple case where just the load is interleaved, because the store group
; would have gaps.
define void @fun0(ptr %data, i64 %n) {
entry:
br label %for.body
for.body:
%i = phi i64 [ 0, %entry ], [ %i.next, %for.body ]
%tmp0 = getelementptr inbounds i32, ptr %data, i64 %i
%tmp1 = load i32, ptr %tmp0, align 4
%tmp2 = add i32 %tmp1, 1
store i32 %tmp2, ptr %tmp0, align 4
%i.next = add nuw nsw i64 %i, 2
%cond = icmp slt i64 %i.next, %n
br i1 %cond, label %for.body, label %for.end
for.end:
ret void
; CHECK: LV: Creating an interleave group with: %tmp1 = load i32, ptr %tmp0, align 4
; CHECK: LV: Found an estimated cost of 3 for VF 4 For instruction: %tmp1 = load i32, ptr %tmp0, align 4
; (vl; vl; vperm)
}
; Interleaving of both load and stores.
define void @fun1(ptr %data, i64 %n) {
entry:
br label %for.body
for.body:
%i = phi i64 [ 0, %entry ], [ %i.next, %for.body ]
%tmp0 = getelementptr inbounds i32, ptr %data, i64 %i
%tmp1 = load i32, ptr %tmp0, align 4
%i_1 = add i64 %i, 1
%tmp2 = getelementptr inbounds i32, ptr %data, i64 %i_1
%tmp3 = load i32, ptr %tmp2, align 4
store i32 %tmp1, ptr %tmp2, align 4
store i32 %tmp3, ptr %tmp0, align 4
%i.next = add nuw nsw i64 %i, 2
%cond = icmp slt i64 %i.next, %n
br i1 %cond, label %for.body, label %for.end
for.end:
ret void
; CHECK: LV: Creating an interleave group with: store i32 %tmp3, ptr %tmp0, align 4
; CHECK: LV: Inserted: store i32 %tmp1, ptr %tmp2, align 4
; CHECK: into the interleave group with store i32 %tmp3, ptr %tmp0, align 4
; CHECK: LV: Creating an interleave group with: %tmp3 = load i32, ptr %tmp2, align 4
; CHECK: LV: Inserted: %tmp1 = load i32, ptr %tmp0, align 4
; CHECK: into the interleave group with %tmp3 = load i32, ptr %tmp2, align 4
; CHECK: LV: Found an estimated cost of 4 for VF 4 For instruction: %tmp1 = load i32, ptr %tmp0, align 4
; CHECK: LV: Found an estimated cost of 0 for VF 4 For instruction: %tmp3 = load i32, ptr %tmp2, align 4
; (vl; vl; vperm, vpkg)
; CHECK: LV: Found an estimated cost of 0 for VF 4 For instruction: store i32 %tmp1, ptr %tmp2, align 4
; CHECK: LV: Found an estimated cost of 4 for VF 4 For instruction: store i32 %tmp3, ptr %tmp0, align 4
; (vmrlf; vmrhf; vst; vst)
}
|
