| # NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py |
| # RUN: llc -march=amdgcn -mcpu=tahiti -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX6 %s |
| # RUN: llc -march=amdgcn -mcpu=hawaii -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX7 %s |
| # RUN: llc -march=amdgcn -mcpu=hawaii -mattr=+flat-for-global -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX7-FLAT %s |
| # RUN: llc -march=amdgcn -mcpu=gfx900 -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX9 %s |
| # RUN: llc -march=amdgcn -mcpu=gfx1010 -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX10 %s |
| |
| --- |
| |
| name: load_atomic_global_s32_seq_cst |
| legalized: true |
| regBankSelected: true |
| tracksRegLiveness: true |
| |
| body: | |
| bb.0: |
| liveins: $vgpr0_vgpr1 |
| |
| ; GFX6-LABEL: name: load_atomic_global_s32_seq_cst |
| ; GFX6: liveins: $vgpr0_vgpr1 |
| ; GFX6: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX6: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 |
| ; GFX6: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 |
| ; GFX6: [[REG_SEQUENCE:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[S_MOV_B32_1]], %subreg.sub1 |
| ; GFX6: [[S_MOV_B64_:%[0-9]+]]:sreg_64 = S_MOV_B64 0 |
| ; GFX6: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[S_MOV_B64_]], %subreg.sub0_sub1, [[REG_SEQUENCE]], %subreg.sub2_sub3 |
| ; GFX6: [[BUFFER_LOAD_DWORD_ADDR64_:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_ADDR64 [[COPY]], [[REG_SEQUENCE1]], 0, 0, 0, 0, 0, implicit $exec :: (load seq_cst 4, addrspace 1) |
| ; GFX6: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_ADDR64_]] |
| ; GFX7-LABEL: name: load_atomic_global_s32_seq_cst |
| ; GFX7: liveins: $vgpr0_vgpr1 |
| ; GFX7: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX7: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 |
| ; GFX7: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 |
| ; GFX7: [[REG_SEQUENCE:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[S_MOV_B32_1]], %subreg.sub1 |
| ; GFX7: [[S_MOV_B64_:%[0-9]+]]:sreg_64 = S_MOV_B64 0 |
| ; GFX7: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[S_MOV_B64_]], %subreg.sub0_sub1, [[REG_SEQUENCE]], %subreg.sub2_sub3 |
| ; GFX7: [[BUFFER_LOAD_DWORD_ADDR64_:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_ADDR64 [[COPY]], [[REG_SEQUENCE1]], 0, 0, 0, 0, 0, implicit $exec :: (load seq_cst 4, addrspace 1) |
| ; GFX7: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_ADDR64_]] |
| ; GFX7-FLAT-LABEL: name: load_atomic_global_s32_seq_cst |
| ; GFX7-FLAT: liveins: $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[FLAT_LOAD_DWORD:%[0-9]+]]:vgpr_32 = FLAT_LOAD_DWORD [[COPY]], 0, 0, implicit $exec, implicit $flat_scr :: (load seq_cst 4, addrspace 1) |
| ; GFX7-FLAT: $vgpr0 = COPY [[FLAT_LOAD_DWORD]] |
| ; GFX9-LABEL: name: load_atomic_global_s32_seq_cst |
| ; GFX9: liveins: $vgpr0_vgpr1 |
| ; GFX9: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX9: [[GLOBAL_LOAD_DWORD:%[0-9]+]]:vgpr_32 = GLOBAL_LOAD_DWORD [[COPY]], 0, 0, implicit $exec :: (load seq_cst 4, addrspace 1) |
| ; GFX9: $vgpr0 = COPY [[GLOBAL_LOAD_DWORD]] |
| ; GFX10-LABEL: name: load_atomic_global_s32_seq_cst |
| ; GFX10: liveins: $vgpr0_vgpr1 |
| ; GFX10: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX10: [[GLOBAL_LOAD_DWORD:%[0-9]+]]:vgpr_32 = GLOBAL_LOAD_DWORD [[COPY]], 0, 0, implicit $exec :: (load seq_cst 4, addrspace 1) |
| ; GFX10: $vgpr0 = COPY [[GLOBAL_LOAD_DWORD]] |
| %0:vgpr(p1) = COPY $vgpr0_vgpr1 |
| %1:vgpr(s32) = G_LOAD %0 :: (load seq_cst 4, align 4, addrspace 1) |
| $vgpr0 = COPY %1 |
| |
| ... |
| |
| --- |
| |
| name: load_atomic_global_v2s16_seq_cst |
| legalized: true |
| regBankSelected: true |
| tracksRegLiveness: true |
| |
| body: | |
| bb.0: |
| liveins: $vgpr0_vgpr1 |
| |
| ; GFX6-LABEL: name: load_atomic_global_v2s16_seq_cst |
| ; GFX6: liveins: $vgpr0_vgpr1 |
| ; GFX6: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX6: [[LOAD:%[0-9]+]]:vgpr_32(<2 x s16>) = G_LOAD [[COPY]](p1) :: (load seq_cst 4, addrspace 1) |
| ; GFX6: $vgpr0 = COPY [[LOAD]](<2 x s16>) |
| ; GFX7-LABEL: name: load_atomic_global_v2s16_seq_cst |
| ; GFX7: liveins: $vgpr0_vgpr1 |
| ; GFX7: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX7: [[LOAD:%[0-9]+]]:vgpr_32(<2 x s16>) = G_LOAD [[COPY]](p1) :: (load seq_cst 4, addrspace 1) |
| ; GFX7: $vgpr0 = COPY [[LOAD]](<2 x s16>) |
| ; GFX7-FLAT-LABEL: name: load_atomic_global_v2s16_seq_cst |
| ; GFX7-FLAT: liveins: $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[LOAD:%[0-9]+]]:vgpr_32(<2 x s16>) = G_LOAD [[COPY]](p1) :: (load seq_cst 4, addrspace 1) |
| ; GFX7-FLAT: $vgpr0 = COPY [[LOAD]](<2 x s16>) |
| ; GFX9-LABEL: name: load_atomic_global_v2s16_seq_cst |
| ; GFX9: liveins: $vgpr0_vgpr1 |
| ; GFX9: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX9: [[LOAD:%[0-9]+]]:vgpr_32(<2 x s16>) = G_LOAD [[COPY]](p1) :: (load seq_cst 4, addrspace 1) |
| ; GFX9: $vgpr0 = COPY [[LOAD]](<2 x s16>) |
| ; GFX10-LABEL: name: load_atomic_global_v2s16_seq_cst |
| ; GFX10: liveins: $vgpr0_vgpr1 |
| ; GFX10: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX10: [[LOAD:%[0-9]+]]:vgpr_32(<2 x s16>) = G_LOAD [[COPY]](p1) :: (load seq_cst 4, addrspace 1) |
| ; GFX10: $vgpr0 = COPY [[LOAD]](<2 x s16>) |
| %0:vgpr(p1) = COPY $vgpr0_vgpr1 |
| %1:vgpr(<2 x s16>) = G_LOAD %0 :: (load seq_cst 4, align 4, addrspace 1) |
| $vgpr0 = COPY %1 |
| |
| ... |
| |
| --- |
| |
| name: load_atomic_global_p3_seq_cst |
| legalized: true |
| regBankSelected: true |
| tracksRegLiveness: true |
| |
| body: | |
| bb.0: |
| liveins: $vgpr0_vgpr1 |
| |
| ; GFX6-LABEL: name: load_atomic_global_p3_seq_cst |
| ; GFX6: liveins: $vgpr0_vgpr1 |
| ; GFX6: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX6: [[LOAD:%[0-9]+]]:vgpr_32(p3) = G_LOAD [[COPY]](p1) :: (load seq_cst 4, addrspace 1) |
| ; GFX6: $vgpr0 = COPY [[LOAD]](p3) |
| ; GFX7-LABEL: name: load_atomic_global_p3_seq_cst |
| ; GFX7: liveins: $vgpr0_vgpr1 |
| ; GFX7: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX7: [[LOAD:%[0-9]+]]:vgpr_32(p3) = G_LOAD [[COPY]](p1) :: (load seq_cst 4, addrspace 1) |
| ; GFX7: $vgpr0 = COPY [[LOAD]](p3) |
| ; GFX7-FLAT-LABEL: name: load_atomic_global_p3_seq_cst |
| ; GFX7-FLAT: liveins: $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[LOAD:%[0-9]+]]:vgpr_32(p3) = G_LOAD [[COPY]](p1) :: (load seq_cst 4, addrspace 1) |
| ; GFX7-FLAT: $vgpr0 = COPY [[LOAD]](p3) |
| ; GFX9-LABEL: name: load_atomic_global_p3_seq_cst |
| ; GFX9: liveins: $vgpr0_vgpr1 |
| ; GFX9: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX9: [[LOAD:%[0-9]+]]:vgpr_32(p3) = G_LOAD [[COPY]](p1) :: (load seq_cst 4, addrspace 1) |
| ; GFX9: $vgpr0 = COPY [[LOAD]](p3) |
| ; GFX10-LABEL: name: load_atomic_global_p3_seq_cst |
| ; GFX10: liveins: $vgpr0_vgpr1 |
| ; GFX10: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX10: [[LOAD:%[0-9]+]]:vgpr_32(p3) = G_LOAD [[COPY]](p1) :: (load seq_cst 4, addrspace 1) |
| ; GFX10: $vgpr0 = COPY [[LOAD]](p3) |
| %0:vgpr(p1) = COPY $vgpr0_vgpr1 |
| %1:vgpr(p3) = G_LOAD %0 :: (load seq_cst 4, align 4, addrspace 1) |
| $vgpr0 = COPY %1 |
| |
| ... |
| |
| --- |
| |
| name: load_atomic_global_s64_seq_cst |
| legalized: true |
| regBankSelected: true |
| tracksRegLiveness: true |
| |
| body: | |
| bb.0: |
| liveins: $vgpr0_vgpr1 |
| |
| ; GFX6-LABEL: name: load_atomic_global_s64_seq_cst |
| ; GFX6: liveins: $vgpr0_vgpr1 |
| ; GFX6: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX6: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 |
| ; GFX6: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 |
| ; GFX6: [[REG_SEQUENCE:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[S_MOV_B32_1]], %subreg.sub1 |
| ; GFX6: [[S_MOV_B64_:%[0-9]+]]:sreg_64 = S_MOV_B64 0 |
| ; GFX6: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[S_MOV_B64_]], %subreg.sub0_sub1, [[REG_SEQUENCE]], %subreg.sub2_sub3 |
| ; GFX6: [[BUFFER_LOAD_DWORDX2_ADDR64_:%[0-9]+]]:vreg_64 = BUFFER_LOAD_DWORDX2_ADDR64 [[COPY]], [[REG_SEQUENCE1]], 0, 0, 0, 0, 0, implicit $exec :: (load seq_cst 8, addrspace 1) |
| ; GFX6: $vgpr0_vgpr1 = COPY [[BUFFER_LOAD_DWORDX2_ADDR64_]] |
| ; GFX7-LABEL: name: load_atomic_global_s64_seq_cst |
| ; GFX7: liveins: $vgpr0_vgpr1 |
| ; GFX7: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX7: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 |
| ; GFX7: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 |
| ; GFX7: [[REG_SEQUENCE:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[S_MOV_B32_1]], %subreg.sub1 |
| ; GFX7: [[S_MOV_B64_:%[0-9]+]]:sreg_64 = S_MOV_B64 0 |
| ; GFX7: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[S_MOV_B64_]], %subreg.sub0_sub1, [[REG_SEQUENCE]], %subreg.sub2_sub3 |
| ; GFX7: [[BUFFER_LOAD_DWORDX2_ADDR64_:%[0-9]+]]:vreg_64 = BUFFER_LOAD_DWORDX2_ADDR64 [[COPY]], [[REG_SEQUENCE1]], 0, 0, 0, 0, 0, implicit $exec :: (load seq_cst 8, addrspace 1) |
| ; GFX7: $vgpr0_vgpr1 = COPY [[BUFFER_LOAD_DWORDX2_ADDR64_]] |
| ; GFX7-FLAT-LABEL: name: load_atomic_global_s64_seq_cst |
| ; GFX7-FLAT: liveins: $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[FLAT_LOAD_DWORDX2_:%[0-9]+]]:vreg_64 = FLAT_LOAD_DWORDX2 [[COPY]], 0, 0, implicit $exec, implicit $flat_scr :: (load seq_cst 8, addrspace 1) |
| ; GFX7-FLAT: $vgpr0_vgpr1 = COPY [[FLAT_LOAD_DWORDX2_]] |
| ; GFX9-LABEL: name: load_atomic_global_s64_seq_cst |
| ; GFX9: liveins: $vgpr0_vgpr1 |
| ; GFX9: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX9: [[GLOBAL_LOAD_DWORDX2_:%[0-9]+]]:vreg_64 = GLOBAL_LOAD_DWORDX2 [[COPY]], 0, 0, implicit $exec :: (load seq_cst 8, addrspace 1) |
| ; GFX9: $vgpr0_vgpr1 = COPY [[GLOBAL_LOAD_DWORDX2_]] |
| ; GFX10-LABEL: name: load_atomic_global_s64_seq_cst |
| ; GFX10: liveins: $vgpr0_vgpr1 |
| ; GFX10: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX10: [[GLOBAL_LOAD_DWORDX2_:%[0-9]+]]:vreg_64 = GLOBAL_LOAD_DWORDX2 [[COPY]], 0, 0, implicit $exec :: (load seq_cst 8, addrspace 1) |
| ; GFX10: $vgpr0_vgpr1 = COPY [[GLOBAL_LOAD_DWORDX2_]] |
| %0:vgpr(p1) = COPY $vgpr0_vgpr1 |
| %1:vgpr(s64) = G_LOAD %0 :: (load seq_cst 8, align 8, addrspace 1) |
| $vgpr0_vgpr1 = COPY %1 |
| |
| ... |
| |
| --- |
| |
| name: load_atomic_global_v2s32_seq_cst |
| legalized: true |
| regBankSelected: true |
| tracksRegLiveness: true |
| |
| body: | |
| bb.0: |
| liveins: $vgpr0_vgpr1 |
| |
| ; GFX6-LABEL: name: load_atomic_global_v2s32_seq_cst |
| ; GFX6: liveins: $vgpr0_vgpr1 |
| ; GFX6: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX6: [[LOAD:%[0-9]+]]:vreg_64(<2 x s32>) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX6: $vgpr0_vgpr1 = COPY [[LOAD]](<2 x s32>) |
| ; GFX7-LABEL: name: load_atomic_global_v2s32_seq_cst |
| ; GFX7: liveins: $vgpr0_vgpr1 |
| ; GFX7: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX7: [[LOAD:%[0-9]+]]:vreg_64(<2 x s32>) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX7: $vgpr0_vgpr1 = COPY [[LOAD]](<2 x s32>) |
| ; GFX7-FLAT-LABEL: name: load_atomic_global_v2s32_seq_cst |
| ; GFX7-FLAT: liveins: $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[LOAD:%[0-9]+]]:vreg_64(<2 x s32>) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX7-FLAT: $vgpr0_vgpr1 = COPY [[LOAD]](<2 x s32>) |
| ; GFX9-LABEL: name: load_atomic_global_v2s32_seq_cst |
| ; GFX9: liveins: $vgpr0_vgpr1 |
| ; GFX9: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX9: [[LOAD:%[0-9]+]]:vreg_64(<2 x s32>) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX9: $vgpr0_vgpr1 = COPY [[LOAD]](<2 x s32>) |
| ; GFX10-LABEL: name: load_atomic_global_v2s32_seq_cst |
| ; GFX10: liveins: $vgpr0_vgpr1 |
| ; GFX10: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX10: [[LOAD:%[0-9]+]]:vreg_64(<2 x s32>) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX10: $vgpr0_vgpr1 = COPY [[LOAD]](<2 x s32>) |
| %0:vgpr(p1) = COPY $vgpr0_vgpr1 |
| %1:vgpr(<2 x s32>) = G_LOAD %0 :: (load seq_cst 8, align 8, addrspace 1) |
| $vgpr0_vgpr1 = COPY %1 |
| |
| ... |
| |
| --- |
| |
| name: load_atomic_global_v4s16_seq_cst |
| legalized: true |
| regBankSelected: true |
| tracksRegLiveness: true |
| |
| body: | |
| bb.0: |
| liveins: $vgpr0_vgpr1 |
| |
| ; GFX6-LABEL: name: load_atomic_global_v4s16_seq_cst |
| ; GFX6: liveins: $vgpr0_vgpr1 |
| ; GFX6: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX6: [[LOAD:%[0-9]+]]:vreg_64(<4 x s16>) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX6: $vgpr0_vgpr1 = COPY [[LOAD]](<4 x s16>) |
| ; GFX7-LABEL: name: load_atomic_global_v4s16_seq_cst |
| ; GFX7: liveins: $vgpr0_vgpr1 |
| ; GFX7: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX7: [[LOAD:%[0-9]+]]:vreg_64(<4 x s16>) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX7: $vgpr0_vgpr1 = COPY [[LOAD]](<4 x s16>) |
| ; GFX7-FLAT-LABEL: name: load_atomic_global_v4s16_seq_cst |
| ; GFX7-FLAT: liveins: $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[LOAD:%[0-9]+]]:vreg_64(<4 x s16>) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX7-FLAT: $vgpr0_vgpr1 = COPY [[LOAD]](<4 x s16>) |
| ; GFX9-LABEL: name: load_atomic_global_v4s16_seq_cst |
| ; GFX9: liveins: $vgpr0_vgpr1 |
| ; GFX9: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX9: [[LOAD:%[0-9]+]]:vreg_64(<4 x s16>) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX9: $vgpr0_vgpr1 = COPY [[LOAD]](<4 x s16>) |
| ; GFX10-LABEL: name: load_atomic_global_v4s16_seq_cst |
| ; GFX10: liveins: $vgpr0_vgpr1 |
| ; GFX10: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX10: [[LOAD:%[0-9]+]]:vreg_64(<4 x s16>) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX10: $vgpr0_vgpr1 = COPY [[LOAD]](<4 x s16>) |
| %0:vgpr(p1) = COPY $vgpr0_vgpr1 |
| %1:vgpr(<4 x s16>) = G_LOAD %0 :: (load seq_cst 8, align 8, addrspace 1) |
| $vgpr0_vgpr1 = COPY %1 |
| |
| ... |
| |
| --- |
| |
| name: load_atomic_global_p1_seq_cst |
| legalized: true |
| regBankSelected: true |
| tracksRegLiveness: true |
| |
| body: | |
| bb.0: |
| liveins: $vgpr0_vgpr1 |
| |
| ; GFX6-LABEL: name: load_atomic_global_p1_seq_cst |
| ; GFX6: liveins: $vgpr0_vgpr1 |
| ; GFX6: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX6: [[LOAD:%[0-9]+]]:vreg_64(p1) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX6: $vgpr0_vgpr1 = COPY [[LOAD]](p1) |
| ; GFX7-LABEL: name: load_atomic_global_p1_seq_cst |
| ; GFX7: liveins: $vgpr0_vgpr1 |
| ; GFX7: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX7: [[LOAD:%[0-9]+]]:vreg_64(p1) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX7: $vgpr0_vgpr1 = COPY [[LOAD]](p1) |
| ; GFX7-FLAT-LABEL: name: load_atomic_global_p1_seq_cst |
| ; GFX7-FLAT: liveins: $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[LOAD:%[0-9]+]]:vreg_64(p1) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX7-FLAT: $vgpr0_vgpr1 = COPY [[LOAD]](p1) |
| ; GFX9-LABEL: name: load_atomic_global_p1_seq_cst |
| ; GFX9: liveins: $vgpr0_vgpr1 |
| ; GFX9: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX9: [[LOAD:%[0-9]+]]:vreg_64(p1) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX9: $vgpr0_vgpr1 = COPY [[LOAD]](p1) |
| ; GFX10-LABEL: name: load_atomic_global_p1_seq_cst |
| ; GFX10: liveins: $vgpr0_vgpr1 |
| ; GFX10: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX10: [[LOAD:%[0-9]+]]:vreg_64(p1) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX10: $vgpr0_vgpr1 = COPY [[LOAD]](p1) |
| %0:vgpr(p1) = COPY $vgpr0_vgpr1 |
| %1:vgpr(p1) = G_LOAD %0 :: (load seq_cst 8, align 8, addrspace 1) |
| $vgpr0_vgpr1 = COPY %1 |
| |
| ... |
| |
| --- |
| |
| name: load_atomic_global_p0_seq_cst |
| legalized: true |
| regBankSelected: true |
| tracksRegLiveness: true |
| |
| body: | |
| bb.0: |
| liveins: $vgpr0_vgpr1 |
| |
| ; GFX6-LABEL: name: load_atomic_global_p0_seq_cst |
| ; GFX6: liveins: $vgpr0_vgpr1 |
| ; GFX6: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX6: [[LOAD:%[0-9]+]]:vreg_64(p0) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX6: $vgpr0_vgpr1 = COPY [[LOAD]](p0) |
| ; GFX7-LABEL: name: load_atomic_global_p0_seq_cst |
| ; GFX7: liveins: $vgpr0_vgpr1 |
| ; GFX7: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX7: [[LOAD:%[0-9]+]]:vreg_64(p0) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX7: $vgpr0_vgpr1 = COPY [[LOAD]](p0) |
| ; GFX7-FLAT-LABEL: name: load_atomic_global_p0_seq_cst |
| ; GFX7-FLAT: liveins: $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[LOAD:%[0-9]+]]:vreg_64(p0) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX7-FLAT: $vgpr0_vgpr1 = COPY [[LOAD]](p0) |
| ; GFX9-LABEL: name: load_atomic_global_p0_seq_cst |
| ; GFX9: liveins: $vgpr0_vgpr1 |
| ; GFX9: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX9: [[LOAD:%[0-9]+]]:vreg_64(p0) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX9: $vgpr0_vgpr1 = COPY [[LOAD]](p0) |
| ; GFX10-LABEL: name: load_atomic_global_p0_seq_cst |
| ; GFX10: liveins: $vgpr0_vgpr1 |
| ; GFX10: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1 |
| ; GFX10: [[LOAD:%[0-9]+]]:vreg_64(p0) = G_LOAD [[COPY]](p1) :: (load seq_cst 8, addrspace 1) |
| ; GFX10: $vgpr0_vgpr1 = COPY [[LOAD]](p0) |
| %0:vgpr(p1) = COPY $vgpr0_vgpr1 |
| %1:vgpr(p0) = G_LOAD %0 :: (load seq_cst 8, align 8, addrspace 1) |
| $vgpr0_vgpr1 = COPY %1 |
| |
| ... |
| |
| --- |
| |
| name: load_atomic_global_s32_seq_cst_gep_m2048 |
| legalized: true |
| regBankSelected: true |
| tracksRegLiveness: true |
| |
| body: | |
| bb.0: |
| liveins: $vgpr0_vgpr1 |
| |
| ; GFX6-LABEL: name: load_atomic_global_s32_seq_cst_gep_m2048 |
| ; GFX6: liveins: $vgpr0_vgpr1 |
| ; GFX6: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX6: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 4294965248, implicit $exec |
| ; GFX6: [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 -1, implicit $exec |
| ; GFX6: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_MOV_B32_e32_]], %subreg.sub0, [[V_MOV_B32_e32_1]], %subreg.sub1 |
| ; GFX6: [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub0 |
| ; GFX6: [[COPY2:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub0 |
| ; GFX6: [[COPY3:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub1 |
| ; GFX6: [[COPY4:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub1 |
| ; GFX6: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY1]], [[COPY2]], 0, implicit $exec |
| ; GFX6: %14:vgpr_32, dead %16:sreg_64_xexec = V_ADDC_U32_e64 [[COPY3]], [[COPY4]], killed [[V_ADD_CO_U32_e64_1]], 0, implicit $exec |
| ; GFX6: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_ADD_CO_U32_e64_]], %subreg.sub0, %14, %subreg.sub1 |
| ; GFX6: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 |
| ; GFX6: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 |
| ; GFX6: [[REG_SEQUENCE2:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[S_MOV_B32_1]], %subreg.sub1 |
| ; GFX6: [[S_MOV_B64_:%[0-9]+]]:sreg_64 = S_MOV_B64 0 |
| ; GFX6: [[REG_SEQUENCE3:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[S_MOV_B64_]], %subreg.sub0_sub1, [[REG_SEQUENCE2]], %subreg.sub2_sub3 |
| ; GFX6: [[BUFFER_LOAD_DWORD_ADDR64_:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_ADDR64 [[REG_SEQUENCE1]], [[REG_SEQUENCE3]], 0, 0, 0, 0, 0, implicit $exec :: (load seq_cst 4, addrspace 1) |
| ; GFX6: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_ADDR64_]] |
| ; GFX7-LABEL: name: load_atomic_global_s32_seq_cst_gep_m2048 |
| ; GFX7: liveins: $vgpr0_vgpr1 |
| ; GFX7: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX7: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 4294965248, implicit $exec |
| ; GFX7: [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 -1, implicit $exec |
| ; GFX7: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_MOV_B32_e32_]], %subreg.sub0, [[V_MOV_B32_e32_1]], %subreg.sub1 |
| ; GFX7: [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub0 |
| ; GFX7: [[COPY2:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub0 |
| ; GFX7: [[COPY3:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub1 |
| ; GFX7: [[COPY4:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub1 |
| ; GFX7: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY1]], [[COPY2]], 0, implicit $exec |
| ; GFX7: %14:vgpr_32, dead %16:sreg_64_xexec = V_ADDC_U32_e64 [[COPY3]], [[COPY4]], killed [[V_ADD_CO_U32_e64_1]], 0, implicit $exec |
| ; GFX7: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_ADD_CO_U32_e64_]], %subreg.sub0, %14, %subreg.sub1 |
| ; GFX7: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 |
| ; GFX7: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 |
| ; GFX7: [[REG_SEQUENCE2:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[S_MOV_B32_1]], %subreg.sub1 |
| ; GFX7: [[S_MOV_B64_:%[0-9]+]]:sreg_64 = S_MOV_B64 0 |
| ; GFX7: [[REG_SEQUENCE3:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[S_MOV_B64_]], %subreg.sub0_sub1, [[REG_SEQUENCE2]], %subreg.sub2_sub3 |
| ; GFX7: [[BUFFER_LOAD_DWORD_ADDR64_:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_ADDR64 [[REG_SEQUENCE1]], [[REG_SEQUENCE3]], 0, 0, 0, 0, 0, implicit $exec :: (load seq_cst 4, addrspace 1) |
| ; GFX7: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_ADDR64_]] |
| ; GFX7-FLAT-LABEL: name: load_atomic_global_s32_seq_cst_gep_m2048 |
| ; GFX7-FLAT: liveins: $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 4294965248, implicit $exec |
| ; GFX7-FLAT: [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 -1, implicit $exec |
| ; GFX7-FLAT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_MOV_B32_e32_]], %subreg.sub0, [[V_MOV_B32_e32_1]], %subreg.sub1 |
| ; GFX7-FLAT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub0 |
| ; GFX7-FLAT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub0 |
| ; GFX7-FLAT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub1 |
| ; GFX7-FLAT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub1 |
| ; GFX7-FLAT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY1]], [[COPY2]], 0, implicit $exec |
| ; GFX7-FLAT: %9:vgpr_32, dead %11:sreg_64_xexec = V_ADDC_U32_e64 [[COPY3]], [[COPY4]], killed [[V_ADD_CO_U32_e64_1]], 0, implicit $exec |
| ; GFX7-FLAT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_ADD_CO_U32_e64_]], %subreg.sub0, %9, %subreg.sub1 |
| ; GFX7-FLAT: [[FLAT_LOAD_DWORD:%[0-9]+]]:vgpr_32 = FLAT_LOAD_DWORD [[REG_SEQUENCE1]], 0, 0, implicit $exec, implicit $flat_scr :: (load seq_cst 4, addrspace 1) |
| ; GFX7-FLAT: $vgpr0 = COPY [[FLAT_LOAD_DWORD]] |
| ; GFX9-LABEL: name: load_atomic_global_s32_seq_cst_gep_m2048 |
| ; GFX9: liveins: $vgpr0_vgpr1 |
| ; GFX9: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX9: [[GLOBAL_LOAD_DWORD:%[0-9]+]]:vgpr_32 = GLOBAL_LOAD_DWORD [[COPY]], -2048, 0, implicit $exec :: (load seq_cst 4, addrspace 1) |
| ; GFX9: $vgpr0 = COPY [[GLOBAL_LOAD_DWORD]] |
| ; GFX10-LABEL: name: load_atomic_global_s32_seq_cst_gep_m2048 |
| ; GFX10: liveins: $vgpr0_vgpr1 |
| ; GFX10: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX10: [[GLOBAL_LOAD_DWORD:%[0-9]+]]:vgpr_32 = GLOBAL_LOAD_DWORD [[COPY]], -2048, 0, implicit $exec :: (load seq_cst 4, addrspace 1) |
| ; GFX10: $vgpr0 = COPY [[GLOBAL_LOAD_DWORD]] |
| %0:vgpr(p1) = COPY $vgpr0_vgpr1 |
| %1:vgpr(s64) = G_CONSTANT i64 -2048 |
| %2:vgpr(p1) = G_PTR_ADD %0, %1 |
| %3:vgpr(s32) = G_LOAD %2 :: (load seq_cst 4, align 4, addrspace 1) |
| $vgpr0 = COPY %3 |
| |
| ... |
| |
| --- |
| |
| name: load_atomic_global_s32_seq_cst_gep_4095 |
| legalized: true |
| regBankSelected: true |
| tracksRegLiveness: true |
| |
| body: | |
| bb.0: |
| liveins: $vgpr0_vgpr1 |
| |
| ; GFX6-LABEL: name: load_atomic_global_s32_seq_cst_gep_4095 |
| ; GFX6: liveins: $vgpr0_vgpr1 |
| ; GFX6: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX6: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 |
| ; GFX6: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 |
| ; GFX6: [[REG_SEQUENCE:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[S_MOV_B32_1]], %subreg.sub1 |
| ; GFX6: [[S_MOV_B64_:%[0-9]+]]:sreg_64 = S_MOV_B64 0 |
| ; GFX6: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[S_MOV_B64_]], %subreg.sub0_sub1, [[REG_SEQUENCE]], %subreg.sub2_sub3 |
| ; GFX6: [[BUFFER_LOAD_DWORD_ADDR64_:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_ADDR64 [[COPY]], [[REG_SEQUENCE1]], 0, 4095, 0, 0, 0, implicit $exec :: (load seq_cst 4, addrspace 1) |
| ; GFX6: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_ADDR64_]] |
| ; GFX7-LABEL: name: load_atomic_global_s32_seq_cst_gep_4095 |
| ; GFX7: liveins: $vgpr0_vgpr1 |
| ; GFX7: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX7: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 |
| ; GFX7: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 |
| ; GFX7: [[REG_SEQUENCE:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[S_MOV_B32_1]], %subreg.sub1 |
| ; GFX7: [[S_MOV_B64_:%[0-9]+]]:sreg_64 = S_MOV_B64 0 |
| ; GFX7: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[S_MOV_B64_]], %subreg.sub0_sub1, [[REG_SEQUENCE]], %subreg.sub2_sub3 |
| ; GFX7: [[BUFFER_LOAD_DWORD_ADDR64_:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_ADDR64 [[COPY]], [[REG_SEQUENCE1]], 0, 4095, 0, 0, 0, implicit $exec :: (load seq_cst 4, addrspace 1) |
| ; GFX7: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_ADDR64_]] |
| ; GFX7-FLAT-LABEL: name: load_atomic_global_s32_seq_cst_gep_4095 |
| ; GFX7-FLAT: liveins: $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 4095, implicit $exec |
| ; GFX7-FLAT: [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec |
| ; GFX7-FLAT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_MOV_B32_e32_]], %subreg.sub0, [[V_MOV_B32_e32_1]], %subreg.sub1 |
| ; GFX7-FLAT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub0 |
| ; GFX7-FLAT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub0 |
| ; GFX7-FLAT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub1 |
| ; GFX7-FLAT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub1 |
| ; GFX7-FLAT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY1]], [[COPY2]], 0, implicit $exec |
| ; GFX7-FLAT: %9:vgpr_32, dead %11:sreg_64_xexec = V_ADDC_U32_e64 [[COPY3]], [[COPY4]], killed [[V_ADD_CO_U32_e64_1]], 0, implicit $exec |
| ; GFX7-FLAT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_ADD_CO_U32_e64_]], %subreg.sub0, %9, %subreg.sub1 |
| ; GFX7-FLAT: [[FLAT_LOAD_DWORD:%[0-9]+]]:vgpr_32 = FLAT_LOAD_DWORD [[REG_SEQUENCE1]], 0, 0, implicit $exec, implicit $flat_scr :: (load seq_cst 4, addrspace 1) |
| ; GFX7-FLAT: $vgpr0 = COPY [[FLAT_LOAD_DWORD]] |
| ; GFX9-LABEL: name: load_atomic_global_s32_seq_cst_gep_4095 |
| ; GFX9: liveins: $vgpr0_vgpr1 |
| ; GFX9: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX9: [[GLOBAL_LOAD_DWORD:%[0-9]+]]:vgpr_32 = GLOBAL_LOAD_DWORD [[COPY]], 4095, 0, implicit $exec :: (load seq_cst 4, addrspace 1) |
| ; GFX9: $vgpr0 = COPY [[GLOBAL_LOAD_DWORD]] |
| ; GFX10-LABEL: name: load_atomic_global_s32_seq_cst_gep_4095 |
| ; GFX10: liveins: $vgpr0_vgpr1 |
| ; GFX10: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX10: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 4095, implicit $exec |
| ; GFX10: [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec |
| ; GFX10: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_MOV_B32_e32_]], %subreg.sub0, [[V_MOV_B32_e32_1]], %subreg.sub1 |
| ; GFX10: [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub0 |
| ; GFX10: [[COPY2:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub0 |
| ; GFX10: [[COPY3:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub1 |
| ; GFX10: [[COPY4:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub1 |
| ; GFX10: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_32_xm0_xexec = V_ADD_CO_U32_e64 [[COPY1]], [[COPY2]], 0, implicit $exec |
| ; GFX10: %9:vgpr_32, dead %11:sreg_32_xm0_xexec = V_ADDC_U32_e64 [[COPY3]], [[COPY4]], killed [[V_ADD_CO_U32_e64_1]], 0, implicit $exec |
| ; GFX10: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_ADD_CO_U32_e64_]], %subreg.sub0, %9, %subreg.sub1 |
| ; GFX10: [[GLOBAL_LOAD_DWORD:%[0-9]+]]:vgpr_32 = GLOBAL_LOAD_DWORD [[REG_SEQUENCE1]], 0, 0, implicit $exec :: (load seq_cst 4, addrspace 1) |
| ; GFX10: $vgpr0 = COPY [[GLOBAL_LOAD_DWORD]] |
| %0:vgpr(p1) = COPY $vgpr0_vgpr1 |
| %1:vgpr(s64) = G_CONSTANT i64 4095 |
| %2:vgpr(p1) = G_PTR_ADD %0, %1 |
| %3:vgpr(s32) = G_LOAD %2 :: (load seq_cst 4, align 4, addrspace 1) |
| $vgpr0 = COPY %3 |
| |
| ... |
| |
| --- |
| |
| name: load_atomic_global_s64_seq_cst_gep_m2048 |
| legalized: true |
| regBankSelected: true |
| tracksRegLiveness: true |
| |
| body: | |
| bb.0: |
| liveins: $vgpr0_vgpr1 |
| |
| ; GFX6-LABEL: name: load_atomic_global_s64_seq_cst_gep_m2048 |
| ; GFX6: liveins: $vgpr0_vgpr1 |
| ; GFX6: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX6: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 4294965248, implicit $exec |
| ; GFX6: [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 -1, implicit $exec |
| ; GFX6: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_MOV_B32_e32_]], %subreg.sub0, [[V_MOV_B32_e32_1]], %subreg.sub1 |
| ; GFX6: [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub0 |
| ; GFX6: [[COPY2:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub0 |
| ; GFX6: [[COPY3:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub1 |
| ; GFX6: [[COPY4:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub1 |
| ; GFX6: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY1]], [[COPY2]], 0, implicit $exec |
| ; GFX6: %14:vgpr_32, dead %16:sreg_64_xexec = V_ADDC_U32_e64 [[COPY3]], [[COPY4]], killed [[V_ADD_CO_U32_e64_1]], 0, implicit $exec |
| ; GFX6: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_ADD_CO_U32_e64_]], %subreg.sub0, %14, %subreg.sub1 |
| ; GFX6: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 |
| ; GFX6: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 |
| ; GFX6: [[REG_SEQUENCE2:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[S_MOV_B32_1]], %subreg.sub1 |
| ; GFX6: [[S_MOV_B64_:%[0-9]+]]:sreg_64 = S_MOV_B64 0 |
| ; GFX6: [[REG_SEQUENCE3:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[S_MOV_B64_]], %subreg.sub0_sub1, [[REG_SEQUENCE2]], %subreg.sub2_sub3 |
| ; GFX6: [[BUFFER_LOAD_DWORDX2_ADDR64_:%[0-9]+]]:vreg_64 = BUFFER_LOAD_DWORDX2_ADDR64 [[REG_SEQUENCE1]], [[REG_SEQUENCE3]], 0, 0, 0, 0, 0, implicit $exec :: (load seq_cst 8, addrspace 1) |
| ; GFX6: $vgpr0_vgpr1 = COPY [[BUFFER_LOAD_DWORDX2_ADDR64_]] |
| ; GFX7-LABEL: name: load_atomic_global_s64_seq_cst_gep_m2048 |
| ; GFX7: liveins: $vgpr0_vgpr1 |
| ; GFX7: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX7: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 4294965248, implicit $exec |
| ; GFX7: [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 -1, implicit $exec |
| ; GFX7: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_MOV_B32_e32_]], %subreg.sub0, [[V_MOV_B32_e32_1]], %subreg.sub1 |
| ; GFX7: [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub0 |
| ; GFX7: [[COPY2:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub0 |
| ; GFX7: [[COPY3:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub1 |
| ; GFX7: [[COPY4:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub1 |
| ; GFX7: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY1]], [[COPY2]], 0, implicit $exec |
| ; GFX7: %14:vgpr_32, dead %16:sreg_64_xexec = V_ADDC_U32_e64 [[COPY3]], [[COPY4]], killed [[V_ADD_CO_U32_e64_1]], 0, implicit $exec |
| ; GFX7: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_ADD_CO_U32_e64_]], %subreg.sub0, %14, %subreg.sub1 |
| ; GFX7: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 |
| ; GFX7: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 61440 |
| ; GFX7: [[REG_SEQUENCE2:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[S_MOV_B32_1]], %subreg.sub1 |
| ; GFX7: [[S_MOV_B64_:%[0-9]+]]:sreg_64 = S_MOV_B64 0 |
| ; GFX7: [[REG_SEQUENCE3:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[S_MOV_B64_]], %subreg.sub0_sub1, [[REG_SEQUENCE2]], %subreg.sub2_sub3 |
| ; GFX7: [[BUFFER_LOAD_DWORDX2_ADDR64_:%[0-9]+]]:vreg_64 = BUFFER_LOAD_DWORDX2_ADDR64 [[REG_SEQUENCE1]], [[REG_SEQUENCE3]], 0, 0, 0, 0, 0, implicit $exec :: (load seq_cst 8, addrspace 1) |
| ; GFX7: $vgpr0_vgpr1 = COPY [[BUFFER_LOAD_DWORDX2_ADDR64_]] |
| ; GFX7-FLAT-LABEL: name: load_atomic_global_s64_seq_cst_gep_m2048 |
| ; GFX7-FLAT: liveins: $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX7-FLAT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 4294965248, implicit $exec |
| ; GFX7-FLAT: [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 -1, implicit $exec |
| ; GFX7-FLAT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_MOV_B32_e32_]], %subreg.sub0, [[V_MOV_B32_e32_1]], %subreg.sub1 |
| ; GFX7-FLAT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub0 |
| ; GFX7-FLAT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub0 |
| ; GFX7-FLAT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub1 |
| ; GFX7-FLAT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub1 |
| ; GFX7-FLAT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY1]], [[COPY2]], 0, implicit $exec |
| ; GFX7-FLAT: %9:vgpr_32, dead %11:sreg_64_xexec = V_ADDC_U32_e64 [[COPY3]], [[COPY4]], killed [[V_ADD_CO_U32_e64_1]], 0, implicit $exec |
| ; GFX7-FLAT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[V_ADD_CO_U32_e64_]], %subreg.sub0, %9, %subreg.sub1 |
| ; GFX7-FLAT: [[FLAT_LOAD_DWORDX2_:%[0-9]+]]:vreg_64 = FLAT_LOAD_DWORDX2 [[REG_SEQUENCE1]], 0, 0, implicit $exec, implicit $flat_scr :: (load seq_cst 8, addrspace 1) |
| ; GFX7-FLAT: $vgpr0_vgpr1 = COPY [[FLAT_LOAD_DWORDX2_]] |
| ; GFX9-LABEL: name: load_atomic_global_s64_seq_cst_gep_m2048 |
| ; GFX9: liveins: $vgpr0_vgpr1 |
| ; GFX9: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX9: [[GLOBAL_LOAD_DWORDX2_:%[0-9]+]]:vreg_64 = GLOBAL_LOAD_DWORDX2 [[COPY]], -2048, 0, implicit $exec :: (load seq_cst 8, addrspace 1) |
| ; GFX9: $vgpr0_vgpr1 = COPY [[GLOBAL_LOAD_DWORDX2_]] |
| ; GFX10-LABEL: name: load_atomic_global_s64_seq_cst_gep_m2048 |
| ; GFX10: liveins: $vgpr0_vgpr1 |
| ; GFX10: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 |
| ; GFX10: [[GLOBAL_LOAD_DWORDX2_:%[0-9]+]]:vreg_64 = GLOBAL_LOAD_DWORDX2 [[COPY]], -2048, 0, implicit $exec :: (load seq_cst 8, addrspace 1) |
| ; GFX10: $vgpr0_vgpr1 = COPY [[GLOBAL_LOAD_DWORDX2_]] |
| %0:vgpr(p1) = COPY $vgpr0_vgpr1 |
| %1:vgpr(s64) = G_CONSTANT i64 -2048 |
| %2:vgpr(p1) = G_PTR_ADD %0, %1 |
| %3:vgpr(s64) = G_LOAD %2 :: (load seq_cst 8, align 8, addrspace 1) |
| $vgpr0_vgpr1 = COPY %3 |
| |
| ... |
