map.ll

; ModuleID = 'map'
target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-v16:16:16-v32:32:32-v64:64:64-v128:128:128-f32:32:32-f64:64:64-n16:32:64"
target triple = "nvptx64-nvidia-cuda"

; Function Attrs: nounwind
define void @map(i64* noalias nocapture %out.ad0, i64 %out.sh0, i64* noalias nocapture readonly %fv0.ad0, i64 %fv0.sh0) #0 {
entry:
  %0 = trunc i64 %out.sh0 to i32
  %1 = tail call i32 @llvm.nvvm.read.ptx.sreg.nctaid.x() #2
  %2 = tail call i32 @llvm.nvvm.read.ptx.sreg.ntid.x() #2
  %3 = mul i32 %2, %1
  %4 = tail call i32 @llvm.nvvm.read.ptx.sreg.tid.x() #2
  %5 = icmp slt i32 %4, %0
  br i1 %5, label %loop1.top, label %loop1.exit

loop1.top:                                        ; preds = %entry, %loop1.top
  %6 = phi i32 [ %11, %loop1.top ], [ %4, %entry ]
  %7 = getelementptr i64* %fv0.ad0, i32 %6
  %8 = load i64* %7
  %9 = add i64 %8, 1
  %10 = getelementptr i64* %out.ad0, i32 %6
  store i64 %9, i64* %10
  %11 = add i32 %6, %3
  %12 = icmp slt i32 %11, %0
  br i1 %12, label %loop1.top, label %loop1.exit

loop1.exit:                                       ; preds = %loop1.top, %entry
  ret void
}

; Function Attrs: nounwind readnone readonly
declare i32 @llvm.nvvm.read.ptx.sreg.nctaid.x() #1

; Function Attrs: nounwind readnone readonly
declare i32 @llvm.nvvm.read.ptx.sreg.ntid.x() #1

; Function Attrs: nounwind readnone readonly
declare i32 @llvm.nvvm.read.ptx.sreg.tid.x() #1

attributes #0 = { nounwind }
attributes #1 = { nounwind readnone readonly }
attributes #2 = { nounwind readonly }

!nvvm.annotations = !{!0}

!0 = metadata !{void (i64*, i64, i64*, i64)* @map, metadata !"kernel", i32 1}