Skip to content

[AMDGPU] Move flat patterns instantiation down. NFC. #94409

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jun 5, 2024
Merged

[AMDGPU] Move flat patterns instantiation down. NFC. #94409

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
223 changes: 111 additions & 112 deletions llvm/lib/Target/AMDGPU/FLATInstructions.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1227,118 +1227,6 @@ class ScratchLoadSVaddrPat_D16 <FLAT_Pseudo inst, SDPatternOperator node, ValueT
(inst $vaddr, $saddr, $offset, 0, $in)
>;

let OtherPredicates = [HasFlatAddressSpace] in {

def : FlatLoadPat <FLAT_LOAD_UBYTE, atomic_load_8_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_UBYTE, atomic_load_8_flat, i16>;
def : FlatLoadPat <FLAT_LOAD_USHORT, atomic_load_16_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_USHORT, atomic_load_16_flat, i16>;
def : FlatLoadPat <FLAT_LOAD_UBYTE, extloadi8_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_UBYTE, zextloadi8_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_SBYTE, sextloadi8_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_UBYTE, extloadi8_flat, i16>;
def : FlatLoadPat <FLAT_LOAD_UBYTE, zextloadi8_flat, i16>;
def : FlatLoadPat <FLAT_LOAD_SBYTE, sextloadi8_flat, i16>;
def : FlatLoadPat <FLAT_LOAD_USHORT, extloadi16_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_USHORT, zextloadi16_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_USHORT, load_flat, i16>;
def : FlatLoadPat <FLAT_LOAD_SSHORT, sextloadi16_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_DWORDX3, load_flat, v3i32>;

def : FlatLoadPat <FLAT_LOAD_DWORD, atomic_load_32_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_DWORDX2, atomic_load_64_flat, i64>;

def : FlatStorePat <FLAT_STORE_BYTE, truncstorei8_flat, i32>;
def : FlatStorePat <FLAT_STORE_SHORT, truncstorei16_flat, i32>;

foreach vt = Reg32Types.types in {
def : FlatLoadPat <FLAT_LOAD_DWORD, load_flat, vt>;
def : FlatStorePat <FLAT_STORE_DWORD, store_flat, vt>;
}

foreach vt = VReg_64.RegTypes in {
def : FlatStorePat <FLAT_STORE_DWORDX2, store_flat, vt>;
def : FlatLoadPat <FLAT_LOAD_DWORDX2, load_flat, vt>;
}

def : FlatStorePat <FLAT_STORE_DWORDX3, store_flat, v3i32>;

foreach vt = VReg_128.RegTypes in {
def : FlatLoadPat <FLAT_LOAD_DWORDX4, load_flat, vt>;
def : FlatStorePat <FLAT_STORE_DWORDX4, store_flat, vt>;
}

def : FlatStorePat <FLAT_STORE_DWORD, atomic_store_32_flat, i32>;
def : FlatStorePat <FLAT_STORE_DWORDX2, atomic_store_64_flat, i64>;
def : FlatStorePat <FLAT_STORE_BYTE, atomic_store_8_flat, i32>;
def : FlatStorePat <FLAT_STORE_BYTE, atomic_store_8_flat, i16>;
def : FlatStorePat <FLAT_STORE_SHORT, atomic_store_16_flat, i32>;
def : FlatStorePat <FLAT_STORE_SHORT, atomic_store_16_flat, i16>;

foreach as = [ "flat", "global" ] in {
defm : FlatAtomicPat <"FLAT_ATOMIC_ADD", "atomic_load_add_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SUB", "atomic_load_sub_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_INC", "atomic_load_uinc_wrap_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_DEC", "atomic_load_udec_wrap_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_AND", "atomic_load_and_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SMAX", "atomic_load_max_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_UMAX", "atomic_load_umax_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SMIN", "atomic_load_min_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_UMIN", "atomic_load_umin_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_OR", "atomic_load_or_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SWAP", "atomic_swap_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_CMPSWAP", "AMDGPUatomic_cmp_swap_"#as, i32, v2i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_XOR", "atomic_load_xor_"#as, i32>;

defm : FlatAtomicPat <"FLAT_ATOMIC_ADD_X2", "atomic_load_add_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SUB_X2", "atomic_load_sub_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_INC_X2", "atomic_load_uinc_wrap_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_DEC_X2", "atomic_load_udec_wrap_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_AND_X2", "atomic_load_and_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SMAX_X2", "atomic_load_max_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_UMAX_X2", "atomic_load_umax_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SMIN_X2", "atomic_load_min_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_UMIN_X2", "atomic_load_umin_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_OR_X2", "atomic_load_or_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SWAP_X2", "atomic_swap_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_CMPSWAP_X2", "AMDGPUatomic_cmp_swap_"#as, i64, v2i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_XOR_X2", "atomic_load_xor_"#as, i64>;
} // end foreach as

let SubtargetPredicate = isGFX12Plus in {
defm : FlatAtomicRtnPatWithAddrSpace<"FLAT_ATOMIC_COND_SUB_U32", "int_amdgcn_atomic_cond_sub_u32", "flat_addrspace", i32 >;

let OtherPredicates = [HasAtomicCSubNoRtnInsts] in
defm : FlatAtomicNoRtnPatWithAddrSpace<"FLAT_ATOMIC_COND_SUB_U32", "int_amdgcn_atomic_cond_sub_u32", "flat_addrspace", i32>;
}

def : FlatStorePat <FLAT_STORE_BYTE, truncstorei8_flat, i16>;
def : FlatStorePat <FLAT_STORE_SHORT, store_flat, i16>;

let OtherPredicates = [HasD16LoadStore] in {
def : FlatStorePat <FLAT_STORE_SHORT_D16_HI, truncstorei16_hi16_flat, i32>;
def : FlatStorePat <FLAT_STORE_BYTE_D16_HI, truncstorei8_hi16_flat, i32>;
}

let OtherPredicates = [D16PreservesUnusedBits] in {
def : FlatLoadPat_D16 <FLAT_LOAD_UBYTE_D16_HI, az_extloadi8_d16_hi_flat, v2i16>;
def : FlatLoadPat_D16 <FLAT_LOAD_UBYTE_D16_HI, az_extloadi8_d16_hi_flat, v2f16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SBYTE_D16_HI, sextloadi8_d16_hi_flat, v2i16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SBYTE_D16_HI, sextloadi8_d16_hi_flat, v2f16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SHORT_D16_HI, load_d16_hi_flat, v2i16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SHORT_D16_HI, load_d16_hi_flat, v2f16>;

def : FlatLoadPat_D16 <FLAT_LOAD_UBYTE_D16, az_extloadi8_d16_lo_flat, v2i16>;
def : FlatLoadPat_D16 <FLAT_LOAD_UBYTE_D16, az_extloadi8_d16_lo_flat, v2f16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SBYTE_D16, sextloadi8_d16_lo_flat, v2i16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SBYTE_D16, sextloadi8_d16_lo_flat, v2f16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SHORT_D16, load_d16_lo_flat, v2i16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SHORT_D16, load_d16_lo_flat, v2f16>;
}

} // End OtherPredicates = [HasFlatAddressSpace]


multiclass GlobalFLATLoadPats<FLAT_Pseudo inst, SDPatternOperator node, ValueType vt> {
def : FlatLoadSignedPat <inst, node, vt> {
let AddedComplexity = 10;
Expand Down Expand Up @@ -1467,6 +1355,117 @@ multiclass ScratchFLATLoadPats_D16<FLAT_Pseudo inst, SDPatternOperator node, Val
}
}

let OtherPredicates = [HasFlatAddressSpace] in {

def : FlatLoadPat <FLAT_LOAD_UBYTE, atomic_load_8_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_UBYTE, atomic_load_8_flat, i16>;
def : FlatLoadPat <FLAT_LOAD_USHORT, atomic_load_16_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_USHORT, atomic_load_16_flat, i16>;
def : FlatLoadPat <FLAT_LOAD_UBYTE, extloadi8_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_UBYTE, zextloadi8_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_SBYTE, sextloadi8_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_UBYTE, extloadi8_flat, i16>;
def : FlatLoadPat <FLAT_LOAD_UBYTE, zextloadi8_flat, i16>;
def : FlatLoadPat <FLAT_LOAD_SBYTE, sextloadi8_flat, i16>;
def : FlatLoadPat <FLAT_LOAD_USHORT, extloadi16_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_USHORT, zextloadi16_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_USHORT, load_flat, i16>;
def : FlatLoadPat <FLAT_LOAD_SSHORT, sextloadi16_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_DWORDX3, load_flat, v3i32>;

def : FlatLoadPat <FLAT_LOAD_DWORD, atomic_load_32_flat, i32>;
def : FlatLoadPat <FLAT_LOAD_DWORDX2, atomic_load_64_flat, i64>;

def : FlatStorePat <FLAT_STORE_BYTE, truncstorei8_flat, i32>;
def : FlatStorePat <FLAT_STORE_SHORT, truncstorei16_flat, i32>;

foreach vt = Reg32Types.types in {
def : FlatLoadPat <FLAT_LOAD_DWORD, load_flat, vt>;
def : FlatStorePat <FLAT_STORE_DWORD, store_flat, vt>;
}

foreach vt = VReg_64.RegTypes in {
def : FlatStorePat <FLAT_STORE_DWORDX2, store_flat, vt>;
def : FlatLoadPat <FLAT_LOAD_DWORDX2, load_flat, vt>;
}

def : FlatStorePat <FLAT_STORE_DWORDX3, store_flat, v3i32>;

foreach vt = VReg_128.RegTypes in {
def : FlatLoadPat <FLAT_LOAD_DWORDX4, load_flat, vt>;
def : FlatStorePat <FLAT_STORE_DWORDX4, store_flat, vt>;
}

def : FlatStorePat <FLAT_STORE_DWORD, atomic_store_32_flat, i32>;
def : FlatStorePat <FLAT_STORE_DWORDX2, atomic_store_64_flat, i64>;
def : FlatStorePat <FLAT_STORE_BYTE, atomic_store_8_flat, i32>;
def : FlatStorePat <FLAT_STORE_BYTE, atomic_store_8_flat, i16>;
def : FlatStorePat <FLAT_STORE_SHORT, atomic_store_16_flat, i32>;
def : FlatStorePat <FLAT_STORE_SHORT, atomic_store_16_flat, i16>;

foreach as = [ "flat", "global" ] in {
defm : FlatAtomicPat <"FLAT_ATOMIC_ADD", "atomic_load_add_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SUB", "atomic_load_sub_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_INC", "atomic_load_uinc_wrap_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_DEC", "atomic_load_udec_wrap_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_AND", "atomic_load_and_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SMAX", "atomic_load_max_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_UMAX", "atomic_load_umax_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SMIN", "atomic_load_min_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_UMIN", "atomic_load_umin_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_OR", "atomic_load_or_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SWAP", "atomic_swap_"#as, i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_CMPSWAP", "AMDGPUatomic_cmp_swap_"#as, i32, v2i32>;
defm : FlatAtomicPat <"FLAT_ATOMIC_XOR", "atomic_load_xor_"#as, i32>;

defm : FlatAtomicPat <"FLAT_ATOMIC_ADD_X2", "atomic_load_add_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SUB_X2", "atomic_load_sub_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_INC_X2", "atomic_load_uinc_wrap_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_DEC_X2", "atomic_load_udec_wrap_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_AND_X2", "atomic_load_and_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SMAX_X2", "atomic_load_max_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_UMAX_X2", "atomic_load_umax_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SMIN_X2", "atomic_load_min_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_UMIN_X2", "atomic_load_umin_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_OR_X2", "atomic_load_or_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_SWAP_X2", "atomic_swap_"#as, i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_CMPSWAP_X2", "AMDGPUatomic_cmp_swap_"#as, i64, v2i64>;
defm : FlatAtomicPat <"FLAT_ATOMIC_XOR_X2", "atomic_load_xor_"#as, i64>;
} // end foreach as

let SubtargetPredicate = isGFX12Plus in {
defm : FlatAtomicRtnPatWithAddrSpace<"FLAT_ATOMIC_COND_SUB_U32", "int_amdgcn_atomic_cond_sub_u32", "flat_addrspace", i32 >;

let OtherPredicates = [HasAtomicCSubNoRtnInsts] in
defm : FlatAtomicNoRtnPatWithAddrSpace<"FLAT_ATOMIC_COND_SUB_U32", "int_amdgcn_atomic_cond_sub_u32", "flat_addrspace", i32>;
}

def : FlatStorePat <FLAT_STORE_BYTE, truncstorei8_flat, i16>;
def : FlatStorePat <FLAT_STORE_SHORT, store_flat, i16>;

let OtherPredicates = [HasD16LoadStore] in {
def : FlatStorePat <FLAT_STORE_SHORT_D16_HI, truncstorei16_hi16_flat, i32>;
def : FlatStorePat <FLAT_STORE_BYTE_D16_HI, truncstorei8_hi16_flat, i32>;
}

let OtherPredicates = [D16PreservesUnusedBits] in {
def : FlatLoadPat_D16 <FLAT_LOAD_UBYTE_D16_HI, az_extloadi8_d16_hi_flat, v2i16>;
def : FlatLoadPat_D16 <FLAT_LOAD_UBYTE_D16_HI, az_extloadi8_d16_hi_flat, v2f16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SBYTE_D16_HI, sextloadi8_d16_hi_flat, v2i16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SBYTE_D16_HI, sextloadi8_d16_hi_flat, v2f16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SHORT_D16_HI, load_d16_hi_flat, v2i16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SHORT_D16_HI, load_d16_hi_flat, v2f16>;

def : FlatLoadPat_D16 <FLAT_LOAD_UBYTE_D16, az_extloadi8_d16_lo_flat, v2i16>;
def : FlatLoadPat_D16 <FLAT_LOAD_UBYTE_D16, az_extloadi8_d16_lo_flat, v2f16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SBYTE_D16, sextloadi8_d16_lo_flat, v2i16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SBYTE_D16, sextloadi8_d16_lo_flat, v2f16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SHORT_D16, load_d16_lo_flat, v2i16>;
def : FlatLoadPat_D16 <FLAT_LOAD_SHORT_D16, load_d16_lo_flat, v2f16>;
}

} // End OtherPredicates = [HasFlatAddressSpace]

let OtherPredicates = [HasFlatGlobalInsts] in {

defm : GlobalFLATLoadPats <GLOBAL_LOAD_UBYTE, atomic_load_8_global, i32>;
Expand Down