1; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-4 -load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT4,ALIGNED,ALL %s 2; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-8 -load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT8,ALIGNED,ALL %s 3; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-8,+unaligned-scratch-access -load-store-vectorizer -S -o - %s | FileCheck -check-prefix=ELT8-UNALIGNED -check-prefix=ALL %s 4; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-16 -load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT16,ALIGNED,ALL %s 5; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-16,+unaligned-scratch-access -load-store-vectorizer -S -o - %s | FileCheck -check-prefix=ELT16-UNALIGNED -check-prefix=ALL %s 6 7target 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" 8 9; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v4i32 10; ALIGNED: store i32 11; ALIGNED: store i32 12; ALIGNED: store i32 13; ALIGNED: store i32 14 15; ELT8-UNALIGNED: store <2 x i32> 16; ELT8-UNALIGNED: store <2 x i32> 17 18; ELT16-UNALIGNED: store <4 x i32> 19define void @merge_private_store_4_vector_elts_loads_v4i32(i32* %out) #0 { 20 %out.gep.1 = getelementptr i32, i32* %out, i32 1 21 %out.gep.2 = getelementptr i32, i32* %out, i32 2 22 %out.gep.3 = getelementptr i32, i32* %out, i32 3 23 24 store i32 9, i32* %out 25 store i32 1, i32* %out.gep.1 26 store i32 23, i32* %out.gep.2 27 store i32 19, i32* %out.gep.3 28 ret void 29} 30 31; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v4i8( 32; ALL: store <4 x i8> 33define void @merge_private_store_4_vector_elts_loads_v4i8(i8* %out) #0 { 34 %out.gep.1 = getelementptr i8, i8* %out, i32 1 35 %out.gep.2 = getelementptr i8, i8* %out, i32 2 36 %out.gep.3 = getelementptr i8, i8* %out, i32 3 37 38 store i8 9, i8* %out, align 4 39 store i8 1, i8* %out.gep.1 40 store i8 23, i8* %out.gep.2 41 store i8 19, i8* %out.gep.3 42 ret void 43} 44 45; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v2i16( 46; ALL: store <2 x i16> 47define void @merge_private_store_4_vector_elts_loads_v2i16(i16* %out) #0 { 48 %out.gep.1 = getelementptr i16, i16* %out, i32 1 49 50 store i16 9, i16* %out, align 4 51 store i16 12, i16* %out.gep.1 52 ret void 53} 54 55attributes #0 = { nounwind } 56