caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
a033bf242f7dccd5546292bf6e554cf2de3ccd58
clang-p2996/llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.struct.ptr.buffer.atomic.add.ll
Krzysztof Drewniak ab37937812 [AMDGPU] Use resource base for buffer instruction MachineMemOperands 
1. Remove the existing code that would encode the constant offsets (if
there were any) on buffer intrinsic operations onto their
`MachineMemOperand`s. As far as I can tell, this use of `offset` has
no substantial impact on the generated code, especially since the same
reasoning is performed by areMemAccessesTriviallyDisjoint().

2. When a buffer resource intrinsic takes a pointer argument as the
base resource/descriptor, place that memory argument in the value
field of the MachineMemOperand attached to that intrinsic.

This is more conservative than what would be produced by more typical
LLVM code using GEP, as the Value (for alias analysis purposes)
corresponding to accessing buffer[0] and buffer[1] is the same.
However, the target-specific analysis of disjoint offsets covers a lot
of the simple usecases.

Despite this limitation, the new buffer intrinsics, combined with
LLVM's existing pointer annotations, allow for non-trivial
optimizations, as seen in the new tests, where marking two buffer
descriptors "noalias" allows merging together loads and stores in a
"load from A, modify loaded value, store to B" sequence, which would
not be possible previously.

Depends on D147547

Reviewed By: arsenm

Differential Revision: https://reviews.llvm.org/D148184
2023-06-05 17:06:57 +00:00

252 lines
18 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck %s
; Natural mapping
define amdgpu_ps float @struct_ptr_buffer_atomic_add_i32__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i32 %val, ptr addrspace(8) inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_ptr_buffer_atomic_add_i32__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_BOTHEN_RTN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call i32 @llvm.amdgcn.struct.ptr.buffer.atomic.add.i32(i32 %val, ptr addrspace(8) %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i32 %ret to float
ret float %cast
}
define amdgpu_ps float @struct_ptr_buffer_atomic_add_i32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i32 %val, ptr addrspace(8) inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_ptr_buffer_atomic_add_i32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_BOTHEN_RTN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call i32 @llvm.amdgcn.struct.ptr.buffer.atomic.add.i32(i32 %val, ptr addrspace(8) %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i32 %ret to float
ret float %cast
}
define amdgpu_ps <2 x float> @struct_ptr_buffer_atomic_add_i64__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i64 %val, ptr addrspace(8) inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_ptr_buffer_atomic_add_i64__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_X2_BOTHEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_ADD_X2_BOTHEN_RTN [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64) on %ir.rsrc, align 1, addrspace 8)
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_X2_BOTHEN_RTN]].sub0
; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_X2_BOTHEN_RTN]].sub1
; CHECK-NEXT: $vgpr0 = COPY [[COPY9]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY10]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%ret = call i64 @llvm.amdgcn.struct.ptr.buffer.atomic.add.i64(i64 %val, ptr addrspace(8) %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i64 %ret to <2 x float>
ret <2 x float> %cast
}
define amdgpu_ps void @struct_ptr_buffer_atomic_add_i64_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i64 %val, ptr addrspace(8) inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_ptr_buffer_atomic_add_i64_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; CHECK-NEXT: BUFFER_ATOMIC_ADD_X2_BOTHEN [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64) on %ir.rsrc, align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0
%ret = call i64 @llvm.amdgcn.struct.ptr.buffer.atomic.add.i64(i64 %val, ptr addrspace(8) %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void
}
; All register operands need legalization
define amdgpu_ps float @struct_ptr_buffer_atomic_add_i32__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, ptr addrspace(8) %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-LABEL: name: struct_ptr_buffer_atomic_add_i32__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset
; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: successors: %bb.2(0x80000000)
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[COPY]]
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[COPY5]]
; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[COPY6]]
; CHECK-NEXT: [[S_MOV_B64_:%[0-9]+]]:sreg_64_xexec = S_MOV_B64 $exec
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2:
; CHECK-NEXT: successors: %bb.3(0x80000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY1]], implicit $exec
; CHECK-NEXT: [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY2]], implicit $exec
; CHECK-NEXT: [[V_READFIRSTLANE_B32_2:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY3]], implicit $exec
; CHECK-NEXT: [[V_READFIRSTLANE_B32_3:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY4]], implicit $exec
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1, [[V_READFIRSTLANE_B32_2]], %subreg.sub2, [[V_READFIRSTLANE_B32_3]], %subreg.sub3
; CHECK-NEXT: [[COPY11:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub0_sub1
; CHECK-NEXT: [[COPY12:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub2_sub3
; CHECK-NEXT: [[COPY13:%[0-9]+]]:sreg_64 = COPY [[REG_SEQUENCE1]].sub0_sub1
; CHECK-NEXT: [[COPY14:%[0-9]+]]:sreg_64 = COPY [[REG_SEQUENCE1]].sub2_sub3
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY13]], [[COPY11]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY14]], [[COPY12]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def $scc
; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY7]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_BOTHEN_RTN [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.5:
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_BOTHEN_RTN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call i32 @llvm.amdgcn.struct.ptr.buffer.atomic.add.i32(i32 %val, ptr addrspace(8) %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i32 %ret to float
ret float %cast
}
; All register operands need legalization
define amdgpu_ps void @struct_ptr_buffer_atomic_add_i32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, ptr addrspace(8) %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-LABEL: name: struct_ptr_buffer_atomic_add_i32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset
; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: successors: %bb.2(0x80000000)
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[COPY]]
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[COPY5]]
; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[COPY6]]
; CHECK-NEXT: [[S_MOV_B64_:%[0-9]+]]:sreg_64_xexec = S_MOV_B64 $exec
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2:
; CHECK-NEXT: successors: %bb.3(0x80000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY1]], implicit $exec
; CHECK-NEXT: [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY2]], implicit $exec
; CHECK-NEXT: [[V_READFIRSTLANE_B32_2:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY3]], implicit $exec
; CHECK-NEXT: [[V_READFIRSTLANE_B32_3:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY4]], implicit $exec
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1, [[V_READFIRSTLANE_B32_2]], %subreg.sub2, [[V_READFIRSTLANE_B32_3]], %subreg.sub3
; CHECK-NEXT: [[COPY11:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub0_sub1
; CHECK-NEXT: [[COPY12:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub2_sub3
; CHECK-NEXT: [[COPY13:%[0-9]+]]:sreg_64 = COPY [[REG_SEQUENCE1]].sub0_sub1
; CHECK-NEXT: [[COPY14:%[0-9]+]]:sreg_64 = COPY [[REG_SEQUENCE1]].sub2_sub3
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY13]], [[COPY11]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY14]], [[COPY12]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def $scc
; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY7]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1
; CHECK-NEXT: BUFFER_ATOMIC_ADD_BOTHEN [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.5:
; CHECK-NEXT: S_ENDPGM 0
%ret = call i32 @llvm.amdgcn.struct.ptr.buffer.atomic.add.i32(i32 %val, ptr addrspace(8) %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void
}
; Natural mapping + slc
define amdgpu_ps float @struct_ptr_buffer_atomic_add_i32__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(i32 %val, ptr addrspace(8) inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_ptr_buffer_atomic_add_i32__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 3, implicit $exec :: (volatile dereferenceable load store (s32) on %ir.rsrc, align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_BOTHEN_RTN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call i32 @llvm.amdgcn.struct.ptr.buffer.atomic.add.i32(i32 %val, ptr addrspace(8) %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
%cast = bitcast i32 %ret to float
ret float %cast
}
declare i32 @llvm.amdgcn.struct.ptr.buffer.atomic.add.i32(i32, ptr addrspace(8), i32, i32, i32, i32 immarg) #0
declare i64 @llvm.amdgcn.struct.ptr.buffer.atomic.add.i64(i64, ptr addrspace(8), i32, i32, i32, i32 immarg) #0
attributes #0 = { nounwind }
Reference in New Issue View Git Blame Copy Permalink