diff options
-rw-r--r-- | llvm/test/CodeGen/AMDGPU/GlobalISel/lds-zero-initializer.ll | 34 | ||||
-rw-r--r-- | llvm/test/CodeGen/AMDGPU/lds-zero-initializer.ll | 24 |
2 files changed, 29 insertions, 29 deletions
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/lds-zero-initializer.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/lds-zero-initializer.ll index b27a7f5..7587aa0 100644 --- a/llvm/test/CodeGen/AMDGPU/GlobalISel/lds-zero-initializer.ll +++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/lds-zero-initializer.ll @@ -1,5 +1,5 @@ -; RUN: llc -mtriple=amdgcn -mcpu=tahiti -global-isel -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck -check-prefixes=GCN,GFX8 %s -; RUN: llc -mtriple=amdgcn -mcpu=tonga -global-isel -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck -check-prefixes=GCN,GFX9 %s +; RUN: llc -mtriple=amdgcn -mcpu=tahiti -global-isel -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck -check-prefixes=GCN,GFX6 %s +; RUN: llc -mtriple=amdgcn -mcpu=tonga -global-isel -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck -check-prefixes=GCN,GFX8 %s ; RUN: not llc -mtriple=amdgcn -mcpu=tahiti -global-isel < %s 2>&1 | FileCheck %s ; RUN: not llc -mtriple=amdgcn -mcpu=tonga -global-isel < %s 2>&1 | FileCheck %s @@ -13,23 +13,23 @@ define amdgpu_kernel void @load_zeroinit_lds_global(ptr addrspace(1) %out, i1 %p ; GCN: bb.1 (%ir-block.0): ; GCN: liveins: $sgpr0_sgpr1 ; GCN: [[COPY:%[0-9]+]]:sreg_64 = COPY $sgpr0_sgpr1 - ; GFX8: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 40 + ; GFX6: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 40 ; GCN: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 target-flags(amdgpu-abs32-lo) @lds - ; GFX8: [[S_ADD_U32_:%[0-9]+]]:sreg_32 = S_ADD_U32 [[S_MOV_B32_1]], [[S_MOV_B32_]], implicit-def dead $scc - ; GFX8: [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM [[COPY]], 9, 0 - ; GFX9: [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM [[COPY]], 36, 0 - ; GFX8: [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[S_ADD_U32_]] + ; GFX6: [[S_ADD_U32_:%[0-9]+]]:sreg_32 = S_ADD_U32 [[S_MOV_B32_1]], [[S_MOV_B32_]], implicit-def dead $scc + ; GFX6: [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM [[COPY]], 9, 0 + ; GFX8: [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM [[COPY]], 36, 0 + ; GFX6: [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[S_ADD_U32_]] ; GCN: $m0 = S_MOV_B32 -1 - ; GFX9: [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_1]] - ; GFX8: [[DS_READ_B32_:%[0-9]+]]:vgpr_32 = DS_READ_B32 [[COPY1]], 0, 0, implicit $m0, implicit $exec - ; GFX9: [[DS_READ_B32_:%[0-9]+]]:vgpr_32 = DS_READ_B32 [[COPY1]], 40, 0, implicit $m0, implicit $exec - ; GFX8: [[S_MOV_B32_2:%[0-9]+]]:sreg_32 = S_MOV_B32 4294967295 - ; GFX8: [[S_MOV_B32_3:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 - ; GFX8: [[REG_SEQUENCE:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[S_MOV_B32_2]], %subreg.sub0, [[S_MOV_B32_3]], %subreg.sub1 - ; GFX8: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[S_LOAD_DWORDX2_IMM]], %subreg.sub0_sub1, [[REG_SEQUENCE]], %subreg.sub2_sub3 - ; GFX8: BUFFER_STORE_DWORD_OFFSET [[DS_READ_B32_]], [[REG_SEQUENCE1]], 0, 0, 0, 0, implicit $exec - ; GFX9: [[COPY2:%[0-9]+]]:vreg_64 = COPY [[S_LOAD_DWORDX2_IMM]] - ; GFX9: FLAT_STORE_DWORD [[COPY2]], [[DS_READ_B32_]], 0, 0, implicit $exec, implicit $flat_scr + ; GFX8: [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_1]] + ; GFX6: [[DS_READ_B32_:%[0-9]+]]:vgpr_32 = DS_READ_B32 [[COPY1]], 0, 0, implicit $m0, implicit $exec + ; GFX8: [[DS_READ_B32_:%[0-9]+]]:vgpr_32 = DS_READ_B32 [[COPY1]], 40, 0, implicit $m0, implicit $exec + ; GFX6: [[S_MOV_B32_2:%[0-9]+]]:sreg_32 = S_MOV_B32 4294967295 + ; GFX6: [[S_MOV_B32_3:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 + ; GFX6: [[REG_SEQUENCE:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[S_MOV_B32_2]], %subreg.sub0, [[S_MOV_B32_3]], %subreg.sub1 + ; GFX6: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[S_LOAD_DWORDX2_IMM]], %subreg.sub0_sub1, [[REG_SEQUENCE]], %subreg.sub2_sub3 + ; GFX6: BUFFER_STORE_DWORD_OFFSET [[DS_READ_B32_]], [[REG_SEQUENCE1]], 0, 0, 0, 0, implicit $exec + ; GFX8: [[COPY2:%[0-9]+]]:vreg_64 = COPY [[S_LOAD_DWORDX2_IMM]] + ; GFX8: FLAT_STORE_DWORD [[COPY2]], [[DS_READ_B32_]], 0, 0, implicit $exec, implicit $flat_scr ; GCN: S_ENDPGM 0 %gep = getelementptr [256 x i32], ptr addrspace(3) @lds, i32 0, i32 10 %ld = load i32, ptr addrspace(3) %gep diff --git a/llvm/test/CodeGen/AMDGPU/lds-zero-initializer.ll b/llvm/test/CodeGen/AMDGPU/lds-zero-initializer.ll index a03109d..e1124f3 100644 --- a/llvm/test/CodeGen/AMDGPU/lds-zero-initializer.ll +++ b/llvm/test/CodeGen/AMDGPU/lds-zero-initializer.ll @@ -1,5 +1,5 @@ -; RUN: llc -mtriple=amdgcn -mcpu=tahiti -stop-after=amdgpu-isel -verify-machineinstrs -o - %s | FileCheck -check-prefixes=GCN,GFX8 %s -; RUN: llc -mtriple=amdgcn -mcpu=tonga -stop-after=amdgpu-isel -verify-machineinstrs -o - %s | FileCheck -check-prefixes=GCN,GFX9 %s +; RUN: llc -mtriple=amdgcn -mcpu=tahiti -stop-after=amdgpu-isel -verify-machineinstrs -o - %s | FileCheck -check-prefixes=GCN,GFX6 %s +; RUN: llc -mtriple=amdgcn -mcpu=tonga -stop-after=amdgpu-isel -verify-machineinstrs -o - %s | FileCheck -check-prefixes=GCN,GFX8 %s ; RUN: not llc -mtriple=amdgcn -mcpu=tahiti < %s 2>&1 | FileCheck %s ; RUN: not llc -mtriple=amdgcn -mcpu=tonga < %s 2>&1 | FileCheck %s @@ -13,19 +13,19 @@ define amdgpu_kernel void @load_zeroinit_lds_global(ptr addrspace(1) %out, i1 %p ; GCN: bb.0 (%ir-block.0): ; GCN: liveins: $sgpr0_sgpr1 ; GCN: [[COPY:%[0-9]+]]:sgpr_64(p4) = COPY $sgpr0_sgpr1 - ; GFX8: [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM [[COPY]](p4), 9, 0 - ; GFX9: [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM [[COPY]](p4), 36, 0 - ; GFX8: [[COPY1:%[0-9]+]]:sreg_32 = COPY [[S_LOAD_DWORDX2_IMM]].sub1 - ; GFX8: [[COPY2:%[0-9]+]]:sreg_32 = COPY [[S_LOAD_DWORDX2_IMM]].sub0 - ; GFX8: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 - ; GFX8: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 -1 - ; GFX8: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE killed [[COPY2]], %subreg.sub0, killed [[COPY1]], %subreg.sub1, killed [[S_MOV_B32_1]], %subreg.sub2, killed [[S_MOV_B32_]], %subreg.sub3 + ; GFX6: [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM [[COPY]](p4), 9, 0 + ; GFX8: [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM [[COPY]](p4), 36, 0 + ; GFX6: [[COPY1:%[0-9]+]]:sreg_32 = COPY [[S_LOAD_DWORDX2_IMM]].sub1 + ; GFX6: [[COPY2:%[0-9]+]]:sreg_32 = COPY [[S_LOAD_DWORDX2_IMM]].sub0 + ; GFX6: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 + ; GFX6: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 -1 + ; GFX6: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE killed [[COPY2]], %subreg.sub0, killed [[COPY1]], %subreg.sub1, killed [[S_MOV_B32_1]], %subreg.sub2, killed [[S_MOV_B32_]], %subreg.sub3 ; GCN: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 target-flags(amdgpu-abs32-lo) @lds, implicit $exec ; GCN: SI_INIT_M0 -1, implicit-def $m0 ; GCN: [[DS_READ_B32_:%[0-9]+]]:vgpr_32 = DS_READ_B32 killed [[V_MOV_B32_e32_]], 40, 0, implicit $m0, implicit $exec - ; GFX9: [[COPY1:%[0-9]+]]:vreg_64 = COPY [[S_LOAD_DWORDX2_IMM]] - ; GFX8: BUFFER_STORE_DWORD_OFFSET killed [[DS_READ_B32_]], killed [[REG_SEQUENCE]], 0, 0, 0, 0, implicit $exec - ; GFX9: FLAT_STORE_DWORD killed [[COPY1]], killed [[DS_READ_B32_]], 0, 0, implicit $exec, implicit $flat_scr + ; GFX8: [[COPY1:%[0-9]+]]:vreg_64 = COPY [[S_LOAD_DWORDX2_IMM]] + ; GFX6: BUFFER_STORE_DWORD_OFFSET killed [[DS_READ_B32_]], killed [[REG_SEQUENCE]], 0, 0, 0, 0, implicit $exec + ; GFX8: FLAT_STORE_DWORD killed [[COPY1]], killed [[DS_READ_B32_]], 0, 0, implicit $exec, implicit $flat_scr ; GCN: S_ENDPGM 0 %gep = getelementptr [256 x i32], ptr addrspace(3) @lds, i32 0, i32 10 %ld = load i32, ptr addrspace(3) %gep |