llvm/test/CodeGen/AMDGPU/reduce-store-width-alignment.ll
Matt Arsenault d706d030af AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel
Currently the default C calling convention functions are treated
the same as compute kernels. Make this explicit so the default
calling convention can be changed to a non-kernel.

Converted with perl -pi -e 's/define void/define amdgpu_kernel void/'
on the relevant test directories (and undoing in one place that actually
wanted a non-kernel).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@298444 91177308-0d34-0410-b5e6-96231b3b80d8
2017-03-21 21:39:51 +00:00

54 lines
2.2 KiB
LLVM

; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
; GCN-LABEL: {{^}}store_v2i32_as_v4i16_align_4:
; GCN: s_load_dwordx2
; GCN: ds_write2_b32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} offset1:1{{$}}
define amdgpu_kernel void @store_v2i32_as_v4i16_align_4(<4 x i16> addrspace(3)* align 4 %out, <2 x i32> %x) #0 {
%x.bc = bitcast <2 x i32> %x to <4 x i16>
store <4 x i16> %x.bc, <4 x i16> addrspace(3)* %out, align 4
ret void
}
; GCN-LABEL: {{^}}store_v4i32_as_v8i16_align_4:
; GCN: s_load_dwordx4
; GCN-DAG: ds_write2_b32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} offset0:2 offset1:3
; GCN-DAG: ds_write2_b32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} offset1:1{{$}}
define amdgpu_kernel void @store_v4i32_as_v8i16_align_4(<8 x i16> addrspace(3)* align 4 %out, <4 x i32> %x) #0 {
%x.bc = bitcast <4 x i32> %x to <8 x i16>
store <8 x i16> %x.bc, <8 x i16> addrspace(3)* %out, align 4
ret void
}
; GCN-LABEL: {{^}}store_v2i32_as_i64_align_4:
; GCN: s_load_dwordx2
; GCN: ds_write2_b32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} offset1:1{{$}}
define amdgpu_kernel void @store_v2i32_as_i64_align_4(<4 x i16> addrspace(3)* align 4 %out, <2 x i32> %x) #0 {
%x.bc = bitcast <2 x i32> %x to <4 x i16>
store <4 x i16> %x.bc, <4 x i16> addrspace(3)* %out, align 4
ret void
}
; GCN-LABEL: {{^}}store_v4i32_as_v2i64_align_4:
; GCN: s_load_dwordx4
; GCN-DAG: ds_write2_b32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} offset0:2 offset1:3
; GCN-DAG: ds_write2_b32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} offset1:1{{$}}
define amdgpu_kernel void @store_v4i32_as_v2i64_align_4(<2 x i64> addrspace(3)* align 4 %out, <4 x i32> %x) #0 {
%x.bc = bitcast <4 x i32> %x to <2 x i64>
store <2 x i64> %x.bc, <2 x i64> addrspace(3)* %out, align 4
ret void
}
; GCN-LABEL: {{^}}store_v4i16_as_v2i32_align_4:
; GCN: buffer_load_ushort
; GCN: buffer_load_ushort
; GCN: buffer_load_ushort
; GCN: buffer_load_ushort
; GCN: ds_write2_b32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} offset1:1{{$}}
define amdgpu_kernel void @store_v4i16_as_v2i32_align_4(<2 x i32> addrspace(3)* align 4 %out, <4 x i16> %x) #0 {
%x.bc = bitcast <4 x i16> %x to <2 x i32>
store <2 x i32> %x.bc, <2 x i32> addrspace(3)* %out, align 4
ret void
}
attributes #0 = { nounwind }