1; RUN: opt -mcpu=cyclone -mtriple=arm64-apple-ios -loop-data-prefetch -max-prefetch-iters-ahead=100 -S < %s | FileCheck %s --check-prefix=LARGE_PREFETCH --check-prefix=ALL
2; RUN: opt -mcpu=cyclone -mtriple=arm64-apple-ios -loop-data-prefetch -S < %s | FileCheck %s --check-prefix=NO_LARGE_PREFETCH --check-prefix=ALL
3; RUN: opt -mcpu=generic -mtriple=arm64-apple-ios -loop-data-prefetch -S < %s | FileCheck %s --check-prefix=NO_LARGE_PREFETCH --check-prefix=ALL
4; RUN: opt -mcpu=cyclone -mtriple=arm64-apple-ios -passes=loop-data-prefetch -max-prefetch-iters-ahead=100 -S < %s | FileCheck %s --check-prefix=LARGE_PREFETCH --check-prefix=ALL
5; RUN: opt -mcpu=cyclone -mtriple=arm64-apple-ios -passes=loop-data-prefetch -S < %s | FileCheck %s --check-prefix=NO_LARGE_PREFETCH --check-prefix=ALL
6; RUN: opt -mcpu=generic -mtriple=arm64-apple-ios -passes=loop-data-prefetch -S < %s | FileCheck %s --check-prefix=NO_LARGE_PREFETCH --check-prefix=ALL
7
8target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-n32:64-S128"
9
10; ALL-LABEL: @small_stride(
11define void @small_stride(double* nocapture %a, double* nocapture readonly %b) {
12entry:
13  br label %for.body
14
15; ALL: for.body:
16for.body:                                         ; preds = %for.body, %entry
17  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
18  %arrayidx = getelementptr inbounds double, double* %b, i64 %indvars.iv
19; ALL-NOT: call void @llvm.prefetch
20  %0 = load double, double* %arrayidx, align 8
21  %add = fadd double %0, 1.000000e+00
22  %arrayidx2 = getelementptr inbounds double, double* %a, i64 %indvars.iv
23  store double %add, double* %arrayidx2, align 8
24  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
25  %exitcond = icmp eq i64 %indvars.iv.next, 1600
26  br i1 %exitcond, label %for.end, label %for.body
27
28; ALL: for.end:
29for.end:                                          ; preds = %for.body
30  ret void
31}
32
33; ALL-LABEL: @large_stride(
34define void @large_stride(double* nocapture %a, double* nocapture readonly %b) {
35entry:
36  br label %for.body
37
38; ALL: for.body:
39for.body:                                         ; preds = %for.body, %entry
40  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
41  %arrayidx = getelementptr inbounds double, double* %b, i64 %indvars.iv
42; LARGE_PREFETCH: call void @llvm.prefetch
43; NO_LARGE_PREFETCH-NOT: call void @llvm.prefetch
44  %0 = load double, double* %arrayidx, align 8
45  %add = fadd double %0, 1.000000e+00
46  %arrayidx2 = getelementptr inbounds double, double* %a, i64 %indvars.iv
47  store double %add, double* %arrayidx2, align 8
48  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 300
49  %exitcond = icmp eq i64 %indvars.iv.next, 160000
50  br i1 %exitcond, label %for.end, label %for.body
51
52; ALL: for.end:
53for.end:                                          ; preds = %for.body
54  ret void
55}
56