File: MachineSink-convergent.ll

package info (click to toggle)
swiftlang 6.0.3-2
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid, trixie
  • size: 2,519,992 kB
  • sloc: cpp: 9,107,863; ansic: 2,040,022; asm: 1,135,751; python: 296,500; objc: 82,456; f90: 60,502; lisp: 34,951; pascal: 19,946; sh: 18,133; perl: 7,482; ml: 4,937; javascript: 4,117; makefile: 3,840; awk: 3,535; xml: 914; fortran: 619; cs: 573; ruby: 573
file content (25 lines) | stat: -rw-r--r-- 692 bytes parent folder | download | duplicates (10) 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
 
; RUN: llc < %s | FileCheck %s
; RUN: %if ptxas %{ llc < %s | %ptxas-verify %}

target triple = "nvptx64-nvidia-cuda"

declare void @llvm.nvvm.barrier0()

; Load a value, then syncthreads.  Branch, and use the loaded value only on one
; side of the branch.  The load shouldn't be sunk beneath the call, because
; syncthreads is modeled as maystore.
define i32 @f(i32 %x, ptr %ptr, i1 %cond) {
Start:
  ; CHECK: ld.u32
  %ptr_val = load i32, ptr %ptr
  ; CHECK: bar.sync
  call void @llvm.nvvm.barrier0()
  br i1 %cond, label %L1, label %L2
L1:
  %ptr_val2 = add i32 %ptr_val, 100
  br label %L2
L2:
  %v4 = phi i32 [ %x, %Start ], [ %ptr_val2, %L1 ]
  %v5 = add i32 %v4, 1000
  ret i32 %v5
}