Skip to content

[AMDGPU] Remove VT helpers isFloatType, isPackedType, simplify isIntType #77987

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 2 commits into from
Jan 16, 2024
Merged

[AMDGPU] Remove VT helpers isFloatType, isPackedType, simplify isIntType #77987

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1fa68d3
[AMDGPU] Remove VT helpers isFloatType, isPackedType, simplify isIntType
rampitec Jan 12, 2024
2a6ca3b
Remove isFP arguments from pseudos in BUF and FLAT.
rampitec Jan 15, 2024
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
10 changes: 4 additions & 6 deletions llvm/lib/Target/AMDGPU/BUFInstructions.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -780,9 +780,8 @@ class MUBUF_AtomicRet_Pseudo<string opName, int addrKind,

multiclass MUBUF_Pseudo_Atomics_NO_RTN <string opName,
RegisterClass vdataClass,
ValueType vdataType,
bit isFP = isFloatType<vdataType>.ret> {
let FPAtomic = isFP in {
ValueType vdataType> {
let FPAtomic = vdataType.isFP in {
def _OFFSET : MUBUF_AtomicNoRet_Pseudo <opName, BUFAddrKind.Offset, vdataClass, 0>,
MUBUFAddr64Table <0, NAME>;
def _ADDR64 : MUBUF_AtomicNoRet_Pseudo <opName, BUFAddrKind.Addr64, vdataClass, 0>,
Expand All @@ -804,9 +803,8 @@ multiclass MUBUF_Pseudo_Atomics_NO_RTN <string opName,
multiclass MUBUF_Pseudo_Atomics_RTN <string opName,
RegisterClass vdataClass,
ValueType vdataType,
SDPatternOperator atomic,
bit isFP = isFloatType<vdataType>.ret> {
let FPAtomic = isFP in {
SDPatternOperator atomic> {
let FPAtomic = vdataType.isFP in {
def _OFFSET_RTN : MUBUF_AtomicRet_Pseudo <opName, BUFAddrKind.Offset, vdataClass, 0,
[(set vdataType:$vdata,
(atomic (MUBUFOffset v4i32:$srsrc, i32:$soffset, i32:$offset),
Expand Down
21 changes: 8 additions & 13 deletions llvm/lib/Target/AMDGPU/FLATInstructions.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -535,7 +535,6 @@ multiclass FLAT_Atomic_Pseudo_NO_RTN<
ValueType vt,
ValueType data_vt = vt,
RegisterClass data_rc = vdst_rc,
bit isFP = isFloatType<data_vt>.ret,
RegisterOperand data_op = getLdStRegisterOperand<data_rc>.ret> {
def "" : FLAT_AtomicNoRet_Pseudo <opName,
(outs),
Expand All @@ -544,7 +543,7 @@ multiclass FLAT_Atomic_Pseudo_NO_RTN<
GlobalSaddrTable<0, opName>,
AtomicNoRet <opName, 0> {
let PseudoInstr = NAME;
let FPAtomic = isFP;
let FPAtomic = data_vt.isFP;
let AddedComplexity = -1; // Prefer global atomics if available
}
}
Expand All @@ -555,15 +554,14 @@ multiclass FLAT_Atomic_Pseudo_RTN<
ValueType vt,
ValueType data_vt = vt,
RegisterClass data_rc = vdst_rc,
bit isFP = isFloatType<data_vt>.ret,
RegisterOperand data_op = getLdStRegisterOperand<data_rc>.ret> {
def _RTN : FLAT_AtomicRet_Pseudo <opName,
(outs getLdStRegisterOperand<vdst_rc>.ret:$vdst),
(ins VReg_64:$vaddr, data_op:$vdata, flat_offset:$offset, CPol_GLC1:$cpol),
" $vdst, $vaddr, $vdata$offset$cpol">,
GlobalSaddrTable<0, opName#"_rtn">,
AtomicNoRet <opName, 1> {
let FPAtomic = isFP;
let FPAtomic = data_vt.isFP;
let AddedComplexity = -1; // Prefer global atomics if available
}
}
Expand All @@ -574,10 +572,9 @@ multiclass FLAT_Atomic_Pseudo<
ValueType vt,
ValueType data_vt = vt,
RegisterClass data_rc = vdst_rc,
bit isFP = isFloatType<data_vt>.ret,
RegisterOperand data_op = getLdStRegisterOperand<data_rc>.ret> {
defm "" : FLAT_Atomic_Pseudo_NO_RTN<opName, vdst_rc, vt, data_vt, data_rc, isFP, data_op>;
defm "" : FLAT_Atomic_Pseudo_RTN<opName, vdst_rc, vt, data_vt, data_rc, isFP, data_op>;
defm "" : FLAT_Atomic_Pseudo_NO_RTN<opName, vdst_rc, vt, data_vt, data_rc, data_op>;
defm "" : FLAT_Atomic_Pseudo_RTN<opName, vdst_rc, vt, data_vt, data_rc, data_op>;
}

multiclass FLAT_Global_Atomic_Pseudo_NO_RTN<
Expand All @@ -586,7 +583,6 @@ multiclass FLAT_Global_Atomic_Pseudo_NO_RTN<
ValueType vt,
ValueType data_vt = vt,
RegisterClass data_rc = vdst_rc,
bit isFP = isFloatType<data_vt>.ret,
RegisterOperand data_op = getLdStRegisterOperand<data_rc>.ret> {

def "" : FLAT_AtomicNoRet_Pseudo <opName,
Expand All @@ -597,7 +593,7 @@ multiclass FLAT_Global_Atomic_Pseudo_NO_RTN<
AtomicNoRet <opName, 0> {
let has_saddr = 1;
let PseudoInstr = NAME;
let FPAtomic = isFP;
let FPAtomic = data_vt.isFP;
}

def _SADDR : FLAT_AtomicNoRet_Pseudo <opName,
Expand All @@ -609,7 +605,7 @@ multiclass FLAT_Global_Atomic_Pseudo_NO_RTN<
let has_saddr = 1;
let enabled_saddr = 1;
let PseudoInstr = NAME#"_SADDR";
let FPAtomic = isFP;
let FPAtomic = data_vt.isFP;
}
}

Expand All @@ -619,7 +615,6 @@ multiclass FLAT_Global_Atomic_Pseudo_RTN<
ValueType vt,
ValueType data_vt = vt,
RegisterClass data_rc = vdst_rc,
bit isFP = isFloatType<data_vt>.ret,
RegisterOperand data_op = getLdStRegisterOperand<data_rc>.ret,
RegisterOperand vdst_op = getLdStRegisterOperand<vdst_rc>.ret> {

Expand All @@ -630,7 +625,7 @@ multiclass FLAT_Global_Atomic_Pseudo_RTN<
GlobalSaddrTable<0, opName#"_rtn">,
AtomicNoRet <opName, 1> {
let has_saddr = 1;
let FPAtomic = isFP;
let FPAtomic = data_vt.isFP;
}

def _SADDR_RTN : FLAT_AtomicRet_Pseudo <opName,
Expand All @@ -642,7 +637,7 @@ multiclass FLAT_Global_Atomic_Pseudo_RTN<
let has_saddr = 1;
let enabled_saddr = 1;
let PseudoInstr = NAME#"_SADDR_RTN";
let FPAtomic = isFP;
let FPAtomic = data_vt.isFP;
}
}

Expand Down
103 changes: 23 additions & 80 deletions llvm/lib/Target/AMDGPU/SIInstrInfo.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -281,56 +281,10 @@ def SIfptrunc_round_downward : SDNode<"AMDGPUISD::FPTRUNC_ROUND_DOWNWARD",
// ValueType helpers
//===----------------------------------------------------------------------===//

// Returns 1 if the source arguments have modifiers, 0 if they do not.
class isFloatType<ValueType SrcVT> {
bit ret = !or(!eq(SrcVT.Value, f16.Value),
!eq(SrcVT.Value, bf16.Value),
!eq(SrcVT.Value, f32.Value),
!eq(SrcVT.Value, f64.Value),
!eq(SrcVT.Value, v2f16.Value),
!eq(SrcVT.Value, v2bf16.Value),
!eq(SrcVT.Value, v4f16.Value),
!eq(SrcVT.Value, v4bf16.Value),
!eq(SrcVT.Value, v8f16.Value),
!eq(SrcVT.Value, v8bf16.Value),
!eq(SrcVT.Value, v16f16.Value),
!eq(SrcVT.Value, v16bf16.Value),
!eq(SrcVT.Value, v2f32.Value),
!eq(SrcVT.Value, v4f32.Value),
!eq(SrcVT.Value, v8f32.Value),
!eq(SrcVT.Value, v2f64.Value),
!eq(SrcVT.Value, v4f64.Value));
}

// XXX - do v2i16 instructions?
class isIntType<ValueType SrcVT> {
bit ret = !or(!eq(SrcVT.Value, i8.Value),
!eq(SrcVT.Value, i16.Value),
!eq(SrcVT.Value, i32.Value),
!eq(SrcVT.Value, i64.Value),
!eq(SrcVT.Value, v4i16.Value),
!eq(SrcVT.Value, v8i16.Value),
!eq(SrcVT.Value, v16i16.Value),
!eq(SrcVT.Value, v2i32.Value),
!eq(SrcVT.Value, v4i32.Value),
!eq(SrcVT.Value, v8i32.Value));
bit ret = !and(SrcVT.isInteger, !ne(SrcVT.Value, i1.Value));
}

class isPackedType<ValueType SrcVT> {
bit ret = !or(!eq(SrcVT.Value, v2i16.Value),
!eq(SrcVT.Value, v2f16.Value),
!eq(SrcVT.Value, v2bf16.Value),
!eq(SrcVT.Value, v4f16.Value),
!eq(SrcVT.Value, v4bf16.Value),
!eq(SrcVT.Value, v2i32.Value),
!eq(SrcVT.Value, v2f32.Value),
!eq(SrcVT.Value, v4i32.Value),
!eq(SrcVT.Value, v4f32.Value),
!eq(SrcVT.Value, v8i32.Value),
!eq(SrcVT.Value, v8f32.Value));
}


//===----------------------------------------------------------------------===//
// PatFrags for global memory operations
//===----------------------------------------------------------------------===//
Expand Down Expand Up @@ -1003,7 +957,7 @@ def ExpSrc3 : RegisterOperand<VGPR_32> {

class SDWASrc<ValueType vt> : RegisterOperand<VS_32> {
let OperandNamespace = "AMDGPU";
string Type = !if(isFloatType<vt>.ret, "FP", "INT");
string Type = !if(vt.isFP, "FP", "INT");
let OperandType = "OPERAND_REG_INLINE_C_"#Type#vt.Size;
let DecoderMethod = "decodeSDWASrc"#vt.Size;
let EncoderMethod = "getSDWASrcEncoding";
Expand Down Expand Up @@ -1499,10 +1453,8 @@ class getSDWADstForVT<ValueType VT> {
// Returns the register class to use for source 0 of VOP[12C]
// instructions for the given VT.
class getVOPSrc0ForVT<ValueType VT, bit IsTrue16, bit IsFake16 = 1> {
bit isFP = isFloatType<VT>.ret;

RegisterOperand ret =
!if(isFP,
!if(VT.isFP,
!if(!eq(VT.Size, 64),
VSrc_f64,
!if(!or(!eq(VT.Value, f16.Value), !eq(VT.Value, bf16.Value)),
Expand Down Expand Up @@ -1562,21 +1514,19 @@ class getVregSrcForVT_t16<ValueType VT, bit IsFake16 = 1> {
}

class getSDWASrcForVT <ValueType VT> {
bit isFP = isFloatType<VT>.ret;
RegisterOperand retFlt = !if(!eq(VT.Size, 16), SDWASrc_f16, SDWASrc_f32);
RegisterOperand retInt = !if(!eq(VT.Size, 16), SDWASrc_i16, SDWASrc_i32);
RegisterOperand ret = !if(isFP, retFlt, retInt);
RegisterOperand ret = !if(VT.isFP, retFlt, retInt);
}

// Returns the register class to use for sources of VOP3 instructions for the
// given VT.
class getVOP3SrcForVT<ValueType VT, bit IsTrue16 = 0> {
bit isFP = isFloatType<VT>.ret;
RegisterOperand ret =
!if(!eq(VT.Size, 128),
VRegSrc_128,
!if(!eq(VT.Size, 64),
!if(isFP,
!if(VT.isFP,
!if(!eq(VT.Value, v2f32.Value),
VSrc_v2f32,
VSrc_f64),
Expand All @@ -1585,7 +1535,7 @@ class getVOP3SrcForVT<ValueType VT, bit IsTrue16 = 0> {
VSrc_b64)),
!if(!eq(VT.Value, i1.Value),
SSrc_i1,
!if(isFP,
!if(VT.isFP,
!if(!or(!eq(VT.Value, f16.Value), !eq(VT.Value, bf16.Value)),
!if(IsTrue16, VSrcT_f16, VSrc_f16),
!if(!or(!eq(VT.Value, v2f16.Value), !eq(VT.Value, v2bf16.Value)),
Expand All @@ -1611,10 +1561,9 @@ class getVOP3SrcForVT<ValueType VT, bit IsTrue16 = 0> {

// Src2 of VOP3 DPP instructions cannot be a literal
class getVOP3DPPSrcForVT<ValueType VT> {
bit isFP = isFloatType<VT>.ret;
RegisterOperand ret =
!if (!eq(VT.Value, i1.Value), SSrc_i1,
!if (isFP,
!if (VT.isFP,
!if (!or(!eq(VT.Value, f16.Value), !eq(VT.Value, bf16.Value)), VCSrc_f16,
!if (!or(!eq(VT.Value, v2f16.Value), !eq(VT.Value, v2bf16.Value)), VCSrc_v2f16, VCSrc_f32)),
!if (!eq(VT.Value, i16.Value), VCSrc_b16,
Expand Down Expand Up @@ -1650,14 +1599,12 @@ class isModifierType<ValueType SrcVT> {

// Return type of input modifiers operand for specified input operand
class getSrcMod <ValueType VT, bit IsTrue16 = 0> {
bit isFP = isFloatType<VT>.ret;
bit isPacked = isPackedType<VT>.ret;
Operand ret = !if(!eq(VT.Size, 64),
!if(isFP, FP64InputMods, Int64InputMods),
!if(VT.isFP, FP64InputMods, Int64InputMods),
!if(!eq(VT.Size, 16),
!if(isFP, !if(IsTrue16, FPT16InputMods, FP16InputMods),
!if(IsTrue16, IntT16InputMods, IntOpSelMods)),
!if(isFP, FP32InputMods, Int32InputMods)));
!if(VT.isFP, !if(IsTrue16, FPT16InputMods, FP16InputMods),
!if(IsTrue16, IntT16InputMods, IntOpSelMods)),
!if(VT.isFP, FP32InputMods, Int32InputMods)));
}

class getOpSelMod <ValueType VT> {
Expand All @@ -1667,14 +1614,12 @@ class getOpSelMod <ValueType VT> {

// Return type of input modifiers operand specified input operand for DPP
class getSrcModDPP <ValueType VT> {
bit isFP = isFloatType<VT>.ret;
Operand ret = !if(isFP, FPVRegInputMods, IntVRegInputMods);
Operand ret = !if(VT.isFP, FPVRegInputMods, IntVRegInputMods);
}

class getSrcModDPP_t16 <ValueType VT> {
bit isFP = isFloatType<VT>.ret;
Operand ret =
!if (isFP,
!if (VT.isFP,
!if (!or(!eq(VT.Value, f16.Value), !eq(VT.Value, bf16.Value)),
FPT16VRegInputMods, FPVRegInputMods),
!if (!eq(VT.Value, i16.Value), IntT16VRegInputMods,
Expand All @@ -1683,10 +1628,8 @@ class getSrcModDPP_t16 <ValueType VT> {

// Return type of input modifiers operand for specified input operand for DPP
class getSrcModVOP3DPP <ValueType VT> {
bit isFP = isFloatType<VT>.ret;
bit isPacked = isPackedType<VT>.ret;
Operand ret =
!if (isFP,
!if (VT.isFP,
!if (!or(!eq(VT.Value, f16.Value), !eq(VT.Value, bf16.Value)),
FP16VCSrcInputMods, FP32VCSrcInputMods),
Int32VCSrcInputMods);
Expand Down Expand Up @@ -2330,26 +2273,26 @@ class VOPProfile <list<ValueType> _ArgVT, bit _EnableClamp = 0> {
field bit HasSrc1 = !ne(Src1VT.Value, untyped.Value);
field bit HasSrc2 = !ne(Src2VT.Value, untyped.Value);

field bit HasSrc0FloatMods = isFloatType<Src0VT>.ret;
field bit HasSrc1FloatMods = isFloatType<Src1VT>.ret;
field bit HasSrc2FloatMods = isFloatType<Src2VT>.ret;
field bit HasSrc0FloatMods = Src0VT.isFP;
field bit HasSrc1FloatMods = Src1VT.isFP;
field bit HasSrc2FloatMods = Src2VT.isFP;

field bit HasSrc0IntMods = isIntType<Src0VT>.ret;
field bit HasSrc1IntMods = isIntType<Src1VT>.ret;
field bit HasSrc2IntMods = isIntType<Src2VT>.ret;

field bit HasClamp = !or(isModifierType<Src0VT>.ret, EnableClamp);
field bit HasSDWAClamp = EmitDst;
field bit HasFPClamp = !and(isFloatType<DstVT>.ret, HasClamp);
field bit HasIntClamp = !if(isFloatType<DstVT>.ret, 0, HasClamp);
field bit HasFPClamp = !and(DstVT.isFP, HasClamp);
field bit HasIntClamp = !if(DstVT.isFP, 0, HasClamp);
field bit HasClampLo = HasClamp;
field bit HasClampHi = !and(isPackedType<DstVT>.ret, HasClamp);
field bit HasClampHi = !and(DstVT.isVector, HasClamp);
field bit HasHigh = 0;

field bit IsPacked = isPackedType<Src0VT>.ret;
field bit IsPacked = Src0VT.isVector;
field bit HasOpSel = IsPacked;
field bit HasOMod = !if(IsVOP3P, 0, isFloatType<DstVT>.ret);
field bit HasSDWAOMod = isFloatType<DstVT>.ret;
field bit HasOMod = !if(IsVOP3P, 0, DstVT.isFP);
field bit HasSDWAOMod = DstVT.isFP;

field bit HasModifiers = !or(isModifierType<Src0VT>.ret,
isModifierType<Src1VT>.ret,
Expand Down
2 changes: 1 addition & 1 deletion llvm/lib/Target/AMDGPU/VOP1Instructions.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,7 +48,7 @@ class VOP1_Pseudo <string opName, VOPProfile P, list<dag> pattern=[], bit VOP1On
let mayStore = 0;
let hasSideEffects = 0;

let ReadsModeReg = !or(isFloatType<P.DstVT>.ret, isFloatType<P.Src0VT>.ret);
let ReadsModeReg = !or(P.DstVT.isFP, P.Src0VT.isFP);

let mayRaiseFPException = ReadsModeReg;

Expand Down
2 changes: 1 addition & 1 deletion llvm/lib/Target/AMDGPU/VOP2Instructions.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -69,7 +69,7 @@ class VOP2_Pseudo <string opName, VOPProfile P, list<dag> pattern=[], string suf
let mayStore = 0;
let hasSideEffects = 0;

let ReadsModeReg = !or(isFloatType<P.DstVT>.ret, isFloatType<P.Src0VT>.ret);
let ReadsModeReg = !or(P.DstVT.isFP, P.Src0VT.isFP);

let mayRaiseFPException = ReadsModeReg;

Expand Down
6 changes: 3 additions & 3 deletions llvm/lib/Target/AMDGPU/VOPCInstructions.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -108,8 +108,8 @@ class VOPC_NoSdst_Profile<list<SchedReadWrite> sched, ValueType vt0,
let InsSDWA = (ins Src0ModSDWA:$src0_modifiers, Src0SDWA:$src0,
Src1ModSDWA:$src1_modifiers, Src1SDWA:$src1,
src0_sel:$src0_sel, src1_sel:$src1_sel);
let AsmVOP3Base = !if(isFloatType<Src0VT>.ret, "$src0_modifiers, $src1_modifiers$clamp",
"$src0, $src1");
let AsmVOP3Base = !if(Src0VT.isFP, "$src0_modifiers, $src1_modifiers$clamp",
"$src0, $src1");
let AsmSDWA9 = "$src0_modifiers, $src1_modifiers $src0_sel $src1_sel";
let EmitDst = 0;
}
Expand Down Expand Up @@ -146,7 +146,7 @@ class VOPC_Pseudo <string opName, VOPC_Profile P, list<dag> pattern=[],
let mayStore = 0;
let hasSideEffects = 0;

let ReadsModeReg = isFloatType<P.Src0VT>.ret;
let ReadsModeReg = P.Src0VT.isFP;

let VALU = 1;
let VOPC = 1;
Expand Down
6 changes: 3 additions & 3 deletions llvm/lib/Target/AMDGPU/VOPInstructions.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -152,7 +152,7 @@ class VOP3_Pseudo <string opName, VOPProfile P, list<dag> pattern = [],
let ClampLo = P.HasClampLo;
let ClampHi = P.HasClampHi;

let ReadsModeReg = !or(isFloatType<P.DstVT>.ret, isFloatType<P.Src0VT>.ret);
let ReadsModeReg = !or(P.DstVT.isFP, P.Src0VT.isFP);

let mayRaiseFPException = ReadsModeReg;
let Uses = !if(ReadsModeReg, [MODE, EXEC], [EXEC]);
Expand Down Expand Up @@ -599,7 +599,7 @@ class VOP_SDWA_Pseudo <string opName, VOPProfile P, list<dag> pattern=[]> :
let VALU = 1;
let SDWA = 1;

let ReadsModeReg = !or(isFloatType<P.DstVT>.ret, isFloatType<P.Src0VT>.ret);
let ReadsModeReg = !or(P.DstVT.isFP, P.Src0VT.isFP);

let mayRaiseFPException = ReadsModeReg;
let Uses = !if(ReadsModeReg, [MODE, EXEC], [EXEC]);
Expand Down Expand Up @@ -811,7 +811,7 @@ class VOP_DPP_Pseudo <string OpName, VOPProfile P, list<dag> pattern=[],
let DPP = 1;
let Size = 8;

let ReadsModeReg = !or(isFloatType<P.DstVT>.ret, isFloatType<P.Src0VT>.ret);
let ReadsModeReg = !or(P.DstVT.isFP, P.Src0VT.isFP);

let mayRaiseFPException = ReadsModeReg;
let Uses = !if(ReadsModeReg, [MODE, EXEC], [EXEC]);
Expand Down