xref: /llvm-project/llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/dont-adjust-globalobj-alignment.ll (revision 263bc7f905c7f354cb80403c0b2754cf8880d1b7)
1; RUN: opt -S -passes=load-store-vectorizer --mcpu=hawaii -mattr=+unaligned-access-mode,+unaligned-scratch-access,+max-private-element-size-16 < %s | FileCheck --match-full-lines %s
2
3target triple = "amdgcn--"
4target 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"
5
6@G = internal addrspace(5) global [8 x i16] undef, align 1
7
8; Verify that the alignment of the global remains at 1, even if we vectorize
9; the stores.
10;
11; CHECK: @G = internal addrspace(5) global [8 x i16] undef, align 1
12
13define void @private_store_2xi16_align2_not_alloca(ptr addrspace(5) %p, ptr addrspace(5) %r) {
14; CHECK: define void @private_store_2xi16_align2_not_alloca(ptr addrspace(5) [[P:%.*]], ptr addrspace(5) [[R:%.*]]) #0 {
15; CHECK-NEXT:    [[GEP0:%.*]] = getelementptr i16, ptr addrspace(5) @G, i32 0
16; CHECK-NEXT:    store <2 x i16> <i16 1, i16 2>, ptr addrspace(5) [[GEP0]], align 1
17; CHECK-NEXT:    ret void
18;
19  %gep0 = getelementptr i16, ptr addrspace(5) @G, i32 0
20  %gep1 = getelementptr i16, ptr addrspace(5) @G, i32 1
21  store i16 1, ptr addrspace(5) %gep0, align 1
22  store i16 2, ptr addrspace(5) %gep1, align 1
23  ret void
24}
25