mirror of
https://github.com/RPCS3/llvm.git
synced 2026-01-31 01:25:19 +01:00
5b7ab74398
The memory location an invariant load is using can never be clobbered by any store, so it's safe to move the load ahead of the store. Differential Revision: https://reviews.llvm.org/D46011 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@330725 91177308-0d34-0410-b5e6-96231b3b80d8
29 lines
1.2 KiB
LLVM
29 lines
1.2 KiB
LLVM
; RUN: opt -mtriple=amdgcn-amd-amdhsa -basicaa -load-store-vectorizer -S -o - %s | FileCheck %s
|
|
|
|
target datalayout = "e-p:32:32-p1:64:64-p2:64:64-p3:32:32-p4:64:64-p5:32:32-p24:64:64-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64"
|
|
|
|
; CHECK-LABEL: @interleave
|
|
; CHECK: load <2 x double>, <2 x double> addrspace(1)* %{{.}}, align 8{{$}}
|
|
; CHECK: store <2 x double> zeroinitializer
|
|
; CHECK: store double %add
|
|
define amdgpu_kernel void @interleave(double addrspace(1)* nocapture %a, double addrspace(1)* nocapture %b, double addrspace(1)* nocapture readonly %c) #0 {
|
|
entry:
|
|
%a.idx.1 = getelementptr inbounds double, double addrspace(1)* %a, i64 1
|
|
%c.idx.1 = getelementptr inbounds double, double addrspace(1)* %c, i64 1
|
|
|
|
%ld.c = load double, double addrspace(1)* %c, align 8
|
|
store double 0.0, double addrspace(1)* %a, align 8 ; Cannot alias invariant load
|
|
|
|
%ld.c.idx.1 = load double, double addrspace(1)* %c.idx.1, align 8, !invariant.load !0
|
|
store double 0.0, double addrspace(1)* %a.idx.1, align 8
|
|
|
|
%add = fadd double %ld.c, %ld.c.idx.1
|
|
store double %add, double addrspace(1)* %b
|
|
|
|
ret void
|
|
}
|
|
|
|
attributes #0 = { nounwind }
|
|
|
|
!0 = !{}
|