xref: /llvm-project/polly/test/IstAstInfo/reduction_modulo_and_loop_reversal_schedule.ll (revision e1f056f692d869708c1898d9d65a69ac5584a0ed)
1; RUN: opt %loadNPMPolly '-passes=polly-import-jscop,print<polly-ast>' -polly-ast-detect-parallel -disable-output < %s | FileCheck %s
2;
3; CHECK:          #pragma known-parallel reduction (+ : MemRef_A)
4; CHECK-NEXT:     for (int c0 = 0; c0 <= 2; c0 += 1) {
5; CHECK-NEXT:       if (c0 == 1) {
6; CHECK-NEXT:         #pragma simd
7; CHECK-NEXT:         for (int c1 = 0; c1 < 2 * n; c1 += 1)
8; CHECK-NEXT:           Stmt_S1(c1);
9; CHECK-NEXT:       } else
10; CHECK-NEXT:         #pragma simd reduction (+ : MemRef_A)
11; CHECK-NEXT:         for (int c1 = (-c0 / 2) + 2; c1 <= 2 * n; c1 += 2)
12; CHECK-NEXT:           Stmt_S0(2 * n - c1);
13; CHECK-NEXT:     }
14;
15;    void rmalrs(int *A, long n) {
16;      for (long i = 0; i < 2 * n; i++)
17; S0:    A[0] += i;
18;      for (long i = 0; i < 2 * n; i++)
19; S1:    A[i + 1] = 1;
20;    }
21;
22target datalayout = "e-m:e-p:32:32-i64:64-v128:64:128-n32-S64"
23
24define void @rmalrs(ptr %A, i32 %n) {
25entry:
26  br label %for.cond
27
28for.cond:                                         ; preds = %for.inc, %entry
29  %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.inc ]
30  %mul = shl nsw i32 %n, 1
31  %cmp = icmp slt i32 %i.0, %mul
32  br i1 %cmp, label %for.body, label %for.end
33
34for.body:                                         ; preds = %for.cond
35  br label %S0
36
37S0:                                               ; preds = %for.body
38  %tmp = load i32, ptr %A, align 4
39  %add = add nsw i32 %tmp, %i.0
40  store i32 %add, ptr %A, align 4
41  br label %for.inc
42
43for.inc:                                          ; preds = %S0
44  %inc = add nsw i32 %i.0, 1
45  br label %for.cond
46
47for.end:                                          ; preds = %for.cond
48  br label %for.cond2
49
50for.cond2:                                        ; preds = %for.inc8, %for.end
51  %i1.0 = phi i32 [ 0, %for.end ], [ %inc9, %for.inc8 ]
52  %mul3 = shl nsw i32 %n, 1
53  %cmp4 = icmp slt i32 %i1.0, %mul3
54  br i1 %cmp4, label %for.body5, label %for.end10
55
56for.body5:                                        ; preds = %for.cond2
57  br label %S1
58
59S1:                                               ; preds = %for.body5
60  %add6 = add nsw i32 %i1.0, 1
61  %arrayidx7 = getelementptr inbounds i32, ptr %A, i32 %add6
62  store i32 1, ptr %arrayidx7, align 4
63  br label %for.inc8
64
65for.inc8:                                         ; preds = %S1
66  %inc9 = add nsw i32 %i1.0, 1
67  br label %for.cond2
68
69for.end10:                                        ; preds = %for.cond2
70  ret void
71}
72
73