1; RUN: opt -S %loadPolly -polly-detect-unprofitable -basicaa -polly-opt-isl -polly-vectorizer=polly -polly-ast -analyze < %s | FileCheck %s
2; RUN: opt -S %loadPolly -polly-detect-unprofitable -basicaa -polly-opt-isl -polly-vectorizer=stripmine -polly-ast -analyze < %s | FileCheck %s
3target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
4target triple = "x86_64-unknown-linux-gnu"
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
11define void @foo() #0 {
12entry:
13  br label %entry.split
14
15entry.split:                                      ; preds = %entry
16  br label %for.cond1.preheader
17
18for.cond1.preheader:                              ; preds = %entry.split, %for.inc28
19  %indvar4 = phi i64 [ 0, %entry.split ], [ %indvar.next5, %for.inc28 ]
20  br label %for.body3
21
22for.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]], [1536 x [1536 x float]]* @C, i64 0, i64 %indvar4, i64 %indvar6
25  store float 0.000000e+00, float* %arrayidx24, align 4
26  br label %for.body8
27
28for.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]], [1536 x [1536 x float]]* @A, i64 0, i64 %indvar4, i64 %indvar
31  %arrayidx20 = getelementptr [1536 x [1536 x float]], [1536 x [1536 x float]]* @B, i64 0, i64 %indvar, i64 %indvar6
32  %0 = load float, float* %arrayidx24, align 4
33  %1 = load float, float* %arrayidx16, align 4
34  %2 = load float, float* %arrayidx20, align 4
35  %mul = fmul float %1, %2
36  %add = fadd float %0, %mul
37  store float %add, float* %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
42for.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
47for.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
52for.end30:                                        ; preds = %for.inc28
53  ret void
54}
55
56attributes #0 = { nounwind uwtable "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf" "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 c1 = 0; c1 <= 1535; c1 += 32)
60; CHECK:   for (int c2 = 0; c2 <= 1535; c2 += 32)
61; CHECK:     for (int c3 = 0; c3 <= 31; c3 += 1)
62; CHECK:       for (int c4 = 0; c4 <= 31; c4 += 4)
63; CHECK:         #pragma simd
64; CHECK:         for (int c5 = c4; c5 <= c4 + 3; c5 += 1)
65; CHECK:           Stmt_for_body3(c1 + c3, c2 + c5);
66; CHECK: #pragma known-parallel
67; CHECK: for (int c1 = 0; c1 <= 1535; c1 += 32)
68; CHECK:   for (int c2 = 0; c2 <= 1535; c2 += 32)
69; CHECK:     for (int c3 = 0; c3 <= 1535; c3 += 32)
70; CHECK:       for (int c4 = 0; c4 <= 31; c4 += 1)
71; CHECK:         for (int c5 = 0; c5 <= 31; c5 += 4)
72; CHECK:           for (int c6 = 0; c6 <= 31; c6 += 1)
73; CHECK:             #pragma simd
74; CHECK:             for (int c7 = c5; c7 <= c5 + 3; c7 += 1)
75; CHECK:               Stmt_for_body8(c1 + c4, c2 + c7, c3 + c6);
76
77!llvm.ident = !{!0}
78
79!0 = !{!"clang version 3.5.0 "}
80