1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2 ; RUN: opt -mtriple=amdgcn-- -mcpu=tonga -passes=load-store-vectorizer -S -o - %s | FileCheck %s
4 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"
6 @lds = internal addrspace(3) global [512 x float] undef, align 4
8 ; The original load has an implicit alignment of 4, and should not
9 ; increase to an align 8 load.
11 define amdgpu_kernel void @load_keep_base_alignment_missing_align(ptr addrspace(1) %out) {
12 ; CHECK-LABEL: @load_keep_base_alignment_missing_align(
13 ; CHECK-NEXT: [[PTR0:%.*]] = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 11
14 ; CHECK-NEXT: [[TMP2:%.*]] = load <2 x float>, ptr addrspace(3) [[PTR0]], align 4
15 ; CHECK-NEXT: [[VAL01:%.*]] = extractelement <2 x float> [[TMP2]], i32 0
16 ; CHECK-NEXT: [[VAL12:%.*]] = extractelement <2 x float> [[TMP2]], i32 1
17 ; CHECK-NEXT: [[ADD:%.*]] = fadd float [[VAL01]], [[VAL12]]
18 ; CHECK-NEXT: store float [[ADD]], ptr addrspace(1) [[OUT:%.*]], align 4
19 ; CHECK-NEXT: ret void
21 %ptr0 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 11
22 %val0 = load float, ptr addrspace(3) %ptr0
24 %ptr1 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 12
25 %val1 = load float, ptr addrspace(3) %ptr1
26 %add = fadd float %val0, %val1
27 store float %add, ptr addrspace(1) %out
31 define amdgpu_kernel void @store_keep_base_alignment_missing_align() {
32 ; CHECK-LABEL: @store_keep_base_alignment_missing_align(
33 ; CHECK-NEXT: [[ARRAYIDX0:%.*]] = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 1
34 ; CHECK-NEXT: store <2 x float> zeroinitializer, ptr addrspace(3) [[ARRAYIDX0]], align 4
35 ; CHECK-NEXT: ret void
37 %arrayidx0 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 1
38 %arrayidx1 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 2
39 store float 0.0, ptr addrspace(3) %arrayidx0
40 store float 0.0, ptr addrspace(3) %arrayidx1