1; RUN: llc -march=amdgcn -mcpu=SI -verify-machineinstrs< %s | FileCheck --check-prefix=SI --check-prefix=CHECK %s 2; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs< %s | FileCheck --check-prefix=CI --check-prefix=CHECK %s 3; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs< %s | FileCheck --check-prefix=CI --check-prefix=CHECK %s 4 5define void @use_gep_address_space([1024 x i32] addrspace(3)* %array) nounwind { 6; CHECK-LABEL: {{^}}use_gep_address_space: 7; CHECK: v_mov_b32_e32 [[PTR:v[0-9]+]], s{{[0-9]+}} 8; CHECK: ds_write_b32 [[PTR]], v{{[0-9]+}} offset:64 9 %p = getelementptr [1024 x i32], [1024 x i32] addrspace(3)* %array, i16 0, i16 16 10 store i32 99, i32 addrspace(3)* %p 11 ret void 12} 13 14define void @use_gep_address_space_large_offset([1024 x i32] addrspace(3)* %array) nounwind { 15; CHECK-LABEL: {{^}}use_gep_address_space_large_offset: 16; The LDS offset will be 65536 bytes, which is larger than the size of LDS on 17; SI, which is why it is being OR'd with the base pointer. 18; SI: s_or_b32 19; CI: s_add_i32 20; CHECK: ds_write_b32 21 %p = getelementptr [1024 x i32], [1024 x i32] addrspace(3)* %array, i16 0, i16 16384 22 store i32 99, i32 addrspace(3)* %p 23 ret void 24} 25 26define void @gep_as_vector_v4(<4 x [1024 x i32] addrspace(3)*> %array) nounwind { 27; CHECK-LABEL: {{^}}gep_as_vector_v4: 28; CHECK: s_add_i32 29; CHECK: s_add_i32 30; CHECK: s_add_i32 31; CHECK: s_add_i32 32 %p = getelementptr [1024 x i32], <4 x [1024 x i32] addrspace(3)*> %array, <4 x i16> zeroinitializer, <4 x i16> <i16 16, i16 16, i16 16, i16 16> 33 %p0 = extractelement <4 x i32 addrspace(3)*> %p, i32 0 34 %p1 = extractelement <4 x i32 addrspace(3)*> %p, i32 1 35 %p2 = extractelement <4 x i32 addrspace(3)*> %p, i32 2 36 %p3 = extractelement <4 x i32 addrspace(3)*> %p, i32 3 37 store i32 99, i32 addrspace(3)* %p0 38 store i32 99, i32 addrspace(3)* %p1 39 store i32 99, i32 addrspace(3)* %p2 40 store i32 99, i32 addrspace(3)* %p3 41 ret void 42} 43 44define void @gep_as_vector_v2(<2 x [1024 x i32] addrspace(3)*> %array) nounwind { 45; CHECK-LABEL: {{^}}gep_as_vector_v2: 46; CHECK: s_add_i32 47; CHECK: s_add_i32 48 %p = getelementptr [1024 x i32], <2 x [1024 x i32] addrspace(3)*> %array, <2 x i16> zeroinitializer, <2 x i16> <i16 16, i16 16> 49 %p0 = extractelement <2 x i32 addrspace(3)*> %p, i32 0 50 %p1 = extractelement <2 x i32 addrspace(3)*> %p, i32 1 51 store i32 99, i32 addrspace(3)* %p0 52 store i32 99, i32 addrspace(3)* %p1 53 ret void 54} 55 56