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
26
27
28
29
|
; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; This is NOT OK to vectorize, as either load may alias either store.
; CHECK: load double
; CHECK: store double 0.000000e+00, ptr addrspace(1) %a,
; CHECK: load double
; CHECK: store double 0.000000e+00, ptr addrspace(1) %a.idx.1
define amdgpu_kernel void @interleave(ptr addrspace(1) nocapture %a, ptr addrspace(1) nocapture %b, ptr addrspace(1) nocapture readonly %c) #0 {
entry:
%a.idx.1 = getelementptr inbounds double, ptr addrspace(1) %a, i64 1
%c.idx.1 = getelementptr inbounds double, ptr addrspace(1) %c, i64 1
%ld.c = load double, ptr addrspace(1) %c, align 8 ; may alias store to %a
store double 0.0, ptr addrspace(1) %a, align 8
%ld.c.idx.1 = load double, ptr addrspace(1) %c.idx.1, align 8 ; may alias store to %a
store double 0.0, ptr addrspace(1) %a.idx.1, align 8
%add = fadd double %ld.c, %ld.c.idx.1
store double %add, ptr addrspace(1) %b
ret void
}
attributes #0 = { nounwind }
|
