1; RUN: llc < %s -march=nvptx -mcpu=sm_20 | FileCheck %s
2
3target 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-v16:16:16-v32:32:32-v64:64:64-v128:128:128-n16:32:64"
4target triple = "nvptx64-nvidia-cuda"
5
6; Function Attrs: nounwind
7; CHECK: .func kernelgen_memcpy
8define ptx_device void @kernelgen_memcpy(i8* nocapture %dst) #0 {
9entry:
10  br i1 undef, label %for.end, label %vector.body
11
12vector.body:                                      ; preds = %vector.body, %entry
13  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
14  %scevgep9 = getelementptr i8, i8* %dst, i64 %index
15  %scevgep910 = bitcast i8* %scevgep9 to <4 x i8>*
16  store <4 x i8> undef, <4 x i8>* %scevgep910, align 1
17  %index.next = add i64 %index, 4
18  %0 = icmp eq i64 undef, %index.next
19  br i1 %0, label %middle.block, label %vector.body
20
21middle.block:                                     ; preds = %vector.body
22  br i1 undef, label %for.end, label %for.body.preheader1
23
24for.body.preheader1:                              ; preds = %middle.block
25  %scevgep2 = getelementptr i8, i8* %dst, i64 0
26  br label %for.body
27
28for.body:                                         ; preds = %for.body, %for.body.preheader1
29  %lsr.iv3 = phi i8* [ %scevgep2, %for.body.preheader1 ], [ %scevgep4, %for.body ]
30  store i8 undef, i8* %lsr.iv3, align 1
31  %scevgep4 = getelementptr i8, i8* %lsr.iv3, i64 1
32  br label %for.body
33
34for.end:                                          ; preds = %middle.block, %entry
35  ret void
36}
37
38attributes #0 = { nounwind "less-precise-fpmad"="false" "frame-pointer"="none" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "no-realign-stack" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
39