llvm-project

1
; RUN: opt -S %loadNPMPolly -aa-pipeline=basic-aa -polly-pattern-matching-based-opts=false -polly-vectorizer=stripmine                         '-passes=polly-opt-isl,print<polly-ast>' -disable-output < %s | FileCheck %s
2
; RUN: opt -S %loadNPMPolly -aa-pipeline=basic-aa -polly-pattern-matching-based-opts=false -polly-vectorizer=stripmine -polly-prevect-width=16 '-passes=polly-opt-isl,print<polly-ast>' -disable-output < %s | FileCheck %s -check-prefix=VEC16
3

4
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
5

6
@C = common global [1536 x [1536 x float]] zeroinitializer, align 16
7
@A = common global [1536 x [1536 x float]] zeroinitializer, align 16
8
@B = common global [1536 x [1536 x float]] zeroinitializer, align 16
9

10
; Function Attrs: nounwind uwtable
11
define void @foo() #0 {
12
entry:
13
  br label %entry.split
14

15
entry.split:                                      ; preds = %entry
16
  br label %for.cond1.preheader
17

18
for.cond1.preheader:                              ; preds = %entry.split, %for.inc28
19
  %indvar4 = phi i64 [ 0, %entry.split ], [ %indvar.next5, %for.inc28 ]
20
  br label %for.body3
21

22
for.body3:                                        ; preds = %for.cond1.preheader, %for.inc25
23
  %indvar6 = phi i64 [ 0, %for.cond1.preheader ], [ %indvar.next7, %for.inc25 ]
24
  %arrayidx24 = getelementptr [1536 x [1536 x float]], ptr @C, i64 0, i64 %indvar4, i64 %indvar6
25
  store float 0.000000e+00, ptr %arrayidx24, align 4
26
  br label %for.body8
27

28
for.body8:                                        ; preds = %for.body3, %for.body8
29
  %indvar = phi i64 [ 0, %for.body3 ], [ %indvar.next, %for.body8 ]
30
  %arrayidx16 = getelementptr [1536 x [1536 x float]], ptr @A, i64 0, i64 %indvar4, i64 %indvar
31
  %arrayidx20 = getelementptr [1536 x [1536 x float]], ptr @B, i64 0, i64 %indvar, i64 %indvar6
32
  %0 = load float, ptr %arrayidx24, align 4
33
  %1 = load float, ptr %arrayidx16, align 4
34
  %2 = load float, ptr %arrayidx20, align 4
35
  %mul = fmul float %1, %2
36
  %add = fadd float %0, %mul
37
  store float %add, ptr %arrayidx24, align 4
38
  %indvar.next = add i64 %indvar, 1
39
  %exitcond = icmp ne i64 %indvar.next, 1536
40
  br i1 %exitcond, label %for.body8, label %for.inc25
41

42
for.inc25:                                        ; preds = %for.body8
43
  %indvar.next7 = add i64 %indvar6, 1
44
  %exitcond8 = icmp ne i64 %indvar.next7, 1536
45
  br i1 %exitcond8, label %for.body3, label %for.inc28
46

47
for.inc28:                                        ; preds = %for.inc25
48
  %indvar.next5 = add i64 %indvar4, 1
49
  %exitcond9 = icmp ne i64 %indvar.next5, 1536
50
  br i1 %exitcond9, label %for.cond1.preheader, label %for.end30
51

52
for.end30:                                        ; preds = %for.inc28
53
  ret void
54
}
55

56
attributes #0 = { nounwind uwtable "less-precise-fpmad"="false" "frame-pointer"="all" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
57

58
; CHECK: #pragma known-parallel
59
; CHECK: for (int c0 = 0; c0 <= 47; c0 += 1)
60
; CHECK:   for (int c1 = 0; c1 <= 47; c1 += 1)
61
; CHECK:     for (int c2 = 0; c2 <= 31; c2 += 1)
62
; CHECK:       for (int c3 = 0; c3 <= 7; c3 += 1)
63
; CHECK:         // SIMD
64
; CHECK:         for (int c4 = 0; c4 <= 3; c4 += 1)
65
; CHECK:           Stmt_for_body3(32 * c0 + c2, 32 * c1 + 4 * c3 + c4);
66
; CHECK: #pragma known-parallel
67
; CHECK: for (int c0 = 0; c0 <= 47; c0 += 1)
68
; CHECK:   for (int c1 = 0; c1 <= 47; c1 += 1)
69
; CHECK:     for (int c2 = 0; c2 <= 47; c2 += 1)
70
; CHECK:       for (int c3 = 0; c3 <= 31; c3 += 1)
71
; CHECK:         for (int c4 = 0; c4 <= 7; c4 += 1)
72
; CHECK:           for (int c5 = 0; c5 <= 31; c5 += 1)
73
; CHECK:             // SIMD
74
; CHECK:             for (int c6 = 0; c6 <= 3; c6 += 1)
75
; CHECK:               Stmt_for_body8(32 * c0 + c3, 32 * c1 + 4 * c4 + c6, 32 * c2 + c5);
76

77
; VEC16: {
78
; VEC16:   #pragma known-parallel
79
; VEC16:   for (int c0 = 0; c0 <= 47; c0 += 1)
80
; VEC16:     for (int c1 = 0; c1 <= 47; c1 += 1)
81
; VEC16:       for (int c2 = 0; c2 <= 31; c2 += 1)
82
; VEC16:         for (int c3 = 0; c3 <= 1; c3 += 1)
83
; VEC16:           // SIMD
84
; VEC16:           for (int c4 = 0; c4 <= 15; c4 += 1)
85
; VEC16:             Stmt_for_body3(32 * c0 + c2, 32 * c1 + 16 * c3 + c4);
86
; VEC16:   #pragma known-parallel
87
; VEC16:   for (int c0 = 0; c0 <= 47; c0 += 1)
88
; VEC16:     for (int c1 = 0; c1 <= 47; c1 += 1)
89
; VEC16:       for (int c2 = 0; c2 <= 47; c2 += 1)
90
; VEC16:         for (int c3 = 0; c3 <= 31; c3 += 1)
91
; VEC16:           for (int c4 = 0; c4 <= 1; c4 += 1)
92
; VEC16:             for (int c5 = 0; c5 <= 31; c5 += 1)
93
; VEC16:               // SIMD
94
; VEC16:               for (int c6 = 0; c6 <= 15; c6 += 1)
95
; VEC16:                 Stmt_for_body8(32 * c0 + c3, 32 * c1 + 16 * c4 + c6, 32 * c2 + c5);
96
; VEC16: }
97

98

99
!llvm.ident = !{!0}
100

101
!0 = !{!"clang version 3.5.0 "}
102