[SPIRV]Added support for extension SPV_INTEL_arbitrary_precision_fixed_point #136085
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Thank you for submitting a Pull Request (PR) to the LLVM Project! This PR will be automatically labeled and the relevant teams will be notified. If you wish to, you can add reviewers by using the "Reviewers" section on this page. If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers. If you have further questions, they may be answered by the LLVM GitHub User Guide. You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums. |
|
@llvm/pr-subscribers-backend-spir-v Author: Aadesh PremKumar (aadeshps-mcw) Changes--Added support for extension SPV_INTEL_arbitrary_precision_fixed_point Patch is 40.09 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/136085.diff 7 Files Affected:
diff --git a/llvm/lib/Target/SPIRV/SPIRVBuiltins.cpp b/llvm/lib/Target/SPIRV/SPIRVBuiltins.cpp
index 16364ab30f280..77f3535bb9e87 100644
--- a/llvm/lib/Target/SPIRV/SPIRVBuiltins.cpp
+++ b/llvm/lib/Target/SPIRV/SPIRVBuiltins.cpp
@@ -697,7 +697,8 @@ static bool buildAtomicStoreInst(const SPIRV::IncomingCall *Call,
MachineIRBuilder &MIRBuilder,
SPIRVGlobalRegistry *GR) {
if (Call->isSpirvOp())
- return buildOpFromWrapper(MIRBuilder, SPIRV::OpAtomicStore, Call, Register(0));
+ return buildOpFromWrapper(MIRBuilder, SPIRV::OpAtomicStore, Call,
+ Register(0));
Register ScopeRegister =
buildConstantIntReg32(SPIRV::Scope::Device, MIRBuilder, GR);
@@ -2282,6 +2283,77 @@ static bool generateBindlessImageINTELInst(const SPIRV::IncomingCall *Call,
return buildBindlessImageINTELInst(Call, Opcode, MIRBuilder, GR);
}
+static bool buildAPFixedPointInst(const SPIRV::IncomingCall *Call,
+ unsigned Opcode, MachineIRBuilder &MIRBuilder,
+ SPIRVGlobalRegistry *GR) {
+ MachineRegisterInfo *MRI = MIRBuilder.getMRI();
+ SmallVector<uint32_t, 1> ImmArgs;
+ Register InputReg = Call->Arguments[0];
+ const Type *RetTy = GR->getTypeForSPIRVType(Call->ReturnType);
+ bool IsSRet = RetTy->isVoidTy();
+
+ if (IsSRet) {
+ const LLT ValTy = MRI->getType(InputReg);
+ Register ActualRetValReg = MRI->createGenericVirtualRegister(ValTy);
+ SPIRVType *InstructionType =
+ GR->getPointeeType(GR->getSPIRVTypeForVReg(InputReg));
+ InputReg = Call->Arguments[1];
+ auto InputType = GR->getTypeForSPIRVType(GR->getSPIRVTypeForVReg(InputReg));
+ Register PtrInputReg;
+ if (InputType->getTypeID() == llvm::Type::TypeID::TypedPointerTyID) {
+ LLT InputLLT = MRI->getType(InputReg);
+ PtrInputReg = MRI->createGenericVirtualRegister(InputLLT);
+ SPIRVType *PtrType =
+ GR->getPointeeType(GR->getSPIRVTypeForVReg(InputReg));
+ MachineMemOperand *MMO1 = MIRBuilder.getMF().getMachineMemOperand(
+ MachinePointerInfo(), MachineMemOperand::MOLoad,
+ InputLLT.getSizeInBytes(), Align(4));
+ MIRBuilder.buildLoad(PtrInputReg, InputReg, *MMO1);
+ MRI->setRegClass(PtrInputReg, &SPIRV::iIDRegClass);
+ GR->assignSPIRVTypeToVReg(PtrType, PtrInputReg, MIRBuilder.getMF());
+ }
+
+ for (unsigned index = 2; index < 7; index++) {
+ ImmArgs.push_back(getConstFromIntrinsic(Call->Arguments[index], MRI));
+ }
+
+ // Emit the instruction
+ auto MIB = MIRBuilder.buildInstr(Opcode)
+ .addDef(ActualRetValReg)
+ .addUse(GR->getSPIRVTypeID(InstructionType));
+ if (PtrInputReg)
+ MIB.addUse(PtrInputReg);
+ else
+ MIB.addUse(InputReg);
+
+ for (uint32_t Imm : ImmArgs)
+ MIB.addImm(Imm);
+ unsigned Size = ValTy.getSizeInBytes();
+ // Store result to the pointer passed in Arg[0]
+ MachineMemOperand *MMO = MIRBuilder.getMF().getMachineMemOperand(
+ MachinePointerInfo(), MachineMemOperand::MOStore, Size, Align(4));
+ MRI->setRegClass(ActualRetValReg, &SPIRV::pIDRegClass);
+ MIRBuilder.buildStore(ActualRetValReg, Call->Arguments[0], *MMO);
+ return true;
+ } else {
+ for (unsigned index = 1; index < 6; index++)
+ ImmArgs.push_back(getConstFromIntrinsic(Call->Arguments[index], MRI));
+
+ return buildOpFromWrapper(MIRBuilder, Opcode, Call,
+ GR->getSPIRVTypeID(Call->ReturnType), ImmArgs);
+ }
+}
+
+static bool generateAPFixedPointInst(const SPIRV::IncomingCall *Call,
+ MachineIRBuilder &MIRBuilder,
+ SPIRVGlobalRegistry *GR) {
+ const SPIRV::DemangledBuiltin *Builtin = Call->Builtin;
+ unsigned Opcode =
+ SPIRV::lookupNativeBuiltin(Builtin->Name, Builtin->Set)->Opcode;
+
+ return buildAPFixedPointInst(Call, Opcode, MIRBuilder, GR);
+}
+
static bool
generateTernaryBitwiseFunctionINTELInst(const SPIRV::IncomingCall *Call,
MachineIRBuilder &MIRBuilder,
@@ -2875,6 +2947,8 @@ std::optional<bool> lowerBuiltin(const StringRef DemangledCall,
return generateExtendedBitOpsInst(Call.get(), MIRBuilder, GR);
case SPIRV::BindlessINTEL:
return generateBindlessImageINTELInst(Call.get(), MIRBuilder, GR);
+ case SPIRV::ArbitraryPrecisionFixedPoint:
+ return generateAPFixedPointInst(Call.get(), MIRBuilder, GR);
case SPIRV::TernaryBitwiseINTEL:
return generateTernaryBitwiseFunctionINTELInst(Call.get(), MIRBuilder, GR);
}
diff --git a/llvm/lib/Target/SPIRV/SPIRVBuiltins.td b/llvm/lib/Target/SPIRV/SPIRVBuiltins.td
index b504e7b04d336..7650b69b48d51 100644
--- a/llvm/lib/Target/SPIRV/SPIRVBuiltins.td
+++ b/llvm/lib/Target/SPIRV/SPIRVBuiltins.td
@@ -67,6 +67,7 @@ def CoopMatr : BuiltinGroup;
def ICarryBorrow : BuiltinGroup;
def ExtendedBitOps : BuiltinGroup;
def BindlessINTEL : BuiltinGroup;
+def ArbitraryPrecisionFixedPoint : BuiltinGroup;
def TernaryBitwiseINTEL : BuiltinGroup;
//===----------------------------------------------------------------------===//
@@ -1132,6 +1133,19 @@ defm : DemangledNativeBuiltin<"clock_read_hilo_device", OpenCL_std, KernelClock,
defm : DemangledNativeBuiltin<"clock_read_hilo_work_group", OpenCL_std, KernelClock, 0, 0, OpReadClockKHR>;
defm : DemangledNativeBuiltin<"clock_read_hilo_sub_group", OpenCL_std, KernelClock, 0, 0, OpReadClockKHR>;
+//SPV_INTEL_arbitrary_precision_fixed_point
+defm : DemangledNativeBuiltin<"__spirv_FixedSqrtINTEL", OpenCL_std, ArbitraryPrecisionFixedPoint, 6 , 8, OpFixedSqrtINTEL>;
+defm : DemangledNativeBuiltin<"__spirv_FixedRecipINTEL", OpenCL_std, ArbitraryPrecisionFixedPoint, 6 , 8, OpFixedRecipINTEL>;
+defm : DemangledNativeBuiltin<"__spirv_FixedRsqrtINTEL", OpenCL_std, ArbitraryPrecisionFixedPoint, 6 , 8, OpFixedRsqrtINTEL>;
+defm : DemangledNativeBuiltin<"__spirv_FixedSinINTEL", OpenCL_std, ArbitraryPrecisionFixedPoint, 6 , 8, OpFixedSinINTEL>;
+defm : DemangledNativeBuiltin<"__spirv_FixedCosINTEL", OpenCL_std, ArbitraryPrecisionFixedPoint, 6 , 8, OpFixedCosINTEL>;
+defm : DemangledNativeBuiltin<"__spirv_FixedSinCosINTEL", OpenCL_std, ArbitraryPrecisionFixedPoint, 6 , 8, OpFixedSinCosINTEL>;
+defm : DemangledNativeBuiltin<"__spirv_FixedSinPiINTEL", OpenCL_std, ArbitraryPrecisionFixedPoint, 6 , 8, OpFixedSinPiINTEL>;
+defm : DemangledNativeBuiltin<"__spirv_FixedCosPiINTEL", OpenCL_std, ArbitraryPrecisionFixedPoint, 6 , 8, OpFixedCosPiINTEL>;
+defm : DemangledNativeBuiltin<"__spirv_FixedSinCosPiINTEL", OpenCL_std, ArbitraryPrecisionFixedPoint, 6 , 8, OpFixedSinCosPiINTEL>;
+defm : DemangledNativeBuiltin<"__spirv_FixedLogINTEL", OpenCL_std, ArbitraryPrecisionFixedPoint, 6 , 8, OpFixedLogINTEL>;
+defm : DemangledNativeBuiltin<"__spirv_FixedExpINTEL", OpenCL_std, ArbitraryPrecisionFixedPoint, 6 , 8, OpFixedExpINTEL>;
+
//===----------------------------------------------------------------------===//
// Class defining an atomic instruction on floating-point numbers.
//
diff --git a/llvm/lib/Target/SPIRV/SPIRVCommandLine.cpp b/llvm/lib/Target/SPIRV/SPIRVCommandLine.cpp
index 86702bbe58f09..6f9c7f7705bfd 100644
--- a/llvm/lib/Target/SPIRV/SPIRVCommandLine.cpp
+++ b/llvm/lib/Target/SPIRV/SPIRVCommandLine.cpp
@@ -94,7 +94,10 @@ static const std::map<std::string, SPIRV::Extension::Extension, std::less<>>
{"SPV_INTEL_fp_max_error",
SPIRV::Extension::Extension::SPV_INTEL_fp_max_error},
{"SPV_INTEL_ternary_bitwise_function",
- SPIRV::Extension::Extension::SPV_INTEL_ternary_bitwise_function}};
+ SPIRV::Extension::Extension::SPV_INTEL_ternary_bitwise_function},
+ {"SPV_INTEL_arbitrary_precision_fixed_point",
+ SPIRV::Extension::Extension::
+ SPV_INTEL_arbitrary_precision_fixed_point}};
bool SPIRVExtensionsParser::parse(cl::Option &O, StringRef ArgName,
StringRef ArgValue,
diff --git a/llvm/lib/Target/SPIRV/SPIRVInstrInfo.td b/llvm/lib/Target/SPIRV/SPIRVInstrInfo.td
index 53064ebb51271..2b40eebbfd7b2 100644
--- a/llvm/lib/Target/SPIRV/SPIRVInstrInfo.td
+++ b/llvm/lib/Target/SPIRV/SPIRVInstrInfo.td
@@ -932,3 +932,27 @@ def OpAliasScopeListDeclINTEL: Op<5913, (outs ID:$res), (ins variable_ops),
// SPV_INTEL_ternary_bitwise_function
def OpBitwiseFunctionINTEL: Op<6242, (outs ID:$res), (ins TYPE:$type, ID:$a, ID:$b, ID:$c, ID:$lut_index),
"$res = OpBitwiseFunctionINTEL $type $a $b $c $lut_index">;
+
+//SPV_INTEL_arbitrary_precision_fixed_point
+def OpFixedSqrtINTEL: Op<5923, (outs ID:$res), (ins TYPE:$result_type, ID:$input, i32imm:$sign, i32imm:$l, i32imm:$rl, i32imm:$q, i32imm:$o),
+ "$res = OpFixedSqrtINTEL $result_type $input $sign $l $rl $q $o">;
+def OpFixedRecipINTEL: Op<5924, (outs ID:$res), (ins TYPE:$result_type, ID:$input, i32imm:$sign, i32imm:$l, i32imm:$rl, i32imm:$q, i32imm:$o),
+ "$res = OpFixedRecipINTEL $result_type $input $sign $l $rl $q $o">;
+def OpFixedRsqrtINTEL: Op<5925, (outs ID:$res), (ins TYPE:$result_type, ID:$input, i32imm:$sign, i32imm:$l, i32imm:$rl, i32imm:$q, i32imm:$o),
+ "$res = OpFixedRsqrtINTEL $result_type $input $sign $l $rl $q $o">;
+def OpFixedSinINTEL: Op<5926, (outs ID:$res), (ins TYPE:$result_type, ID:$input, i32imm:$sign, i32imm:$l, i32imm:$rl, i32imm:$q, i32imm:$o),
+ "$res = OpFixedSinINTEL $result_type $input $sign $l $rl $q $o">;
+def OpFixedCosINTEL: Op<5927, (outs ID:$res), (ins TYPE:$result_type, ID:$input, i32imm:$sign, i32imm:$l, i32imm:$rl, i32imm:$q, i32imm:$o),
+ "$res = OpFixedCosINTEL $result_type $input $sign $l $rl $q $o">;
+def OpFixedSinCosINTEL: Op<5928, (outs ID:$res), (ins TYPE:$result_type, ID:$input, i32imm:$sign, i32imm:$l, i32imm:$rl, i32imm:$q, i32imm:$o),
+ "$res = OpFixedSinCosINTEL $result_type $input $sign $l $rl $q $o">;
+def OpFixedSinPiINTEL: Op<5929, (outs ID:$res), (ins TYPE:$result_type, ID:$input, i32imm:$sign, i32imm:$l, i32imm:$rl, i32imm:$q, i32imm:$o),
+ "$res = OpFixedSinPiINTEL $result_type $input $sign $l $rl $q $o">;
+def OpFixedCosPiINTEL: Op<5930, (outs ID:$res), (ins TYPE:$result_type, ID:$input, i32imm:$sign, i32imm:$l, i32imm:$rl, i32imm:$q, i32imm:$o),
+ "$res = OpFixedCosPiINTEL $result_type $input $sign $l $rl $q $o">;
+def OpFixedSinCosPiINTEL: Op<5931, (outs ID:$res), (ins TYPE:$result_type, ID:$input, i32imm:$sign, i32imm:$l, i32imm:$rl, i32imm:$q, i32imm:$o),
+ "$res = OpFixedSinCosPiINTEL $result_type $input $sign $l $rl $q $o">;
+def OpFixedLogINTEL: Op<5932, (outs ID:$res), (ins TYPE:$result_type, ID:$input, i32imm:$sign, i32imm:$l, i32imm:$rl, i32imm:$q, i32imm:$o),
+ "$res = OpFixedLogINTEL $result_type $input $sign $l $rl $q $o">;
+def OpFixedExpINTEL: Op<5933, (outs ID:$res), (ins TYPE:$result_type, ID:$input, i32imm:$sign, i32imm:$l, i32imm:$rl, i32imm:$q, i32imm:$o),
+ "$res = OpFixedExpINTEL $result_type $input $sign $l $rl $q $o">;
\ No newline at end of file
diff --git a/llvm/lib/Target/SPIRV/SPIRVModuleAnalysis.cpp b/llvm/lib/Target/SPIRV/SPIRVModuleAnalysis.cpp
index b1e5e4328cd32..698f554ab02e6 100644
--- a/llvm/lib/Target/SPIRV/SPIRVModuleAnalysis.cpp
+++ b/llvm/lib/Target/SPIRV/SPIRVModuleAnalysis.cpp
@@ -1516,6 +1516,27 @@ void addInstrRequirements(const MachineInstr &MI,
Reqs.addCapability(SPIRV::Capability::GroupNonUniformRotateKHR);
Reqs.addCapability(SPIRV::Capability::GroupNonUniform);
break;
+ case SPIRV::OpFixedCosINTEL:
+ case SPIRV::OpFixedSinINTEL:
+ case SPIRV::OpFixedCosPiINTEL:
+ case SPIRV::OpFixedSinPiINTEL:
+ case SPIRV::OpFixedExpINTEL:
+ case SPIRV::OpFixedLogINTEL:
+ case SPIRV::OpFixedRecipINTEL:
+ case SPIRV::OpFixedSqrtINTEL:
+ case SPIRV::OpFixedSinCosINTEL:
+ case SPIRV::OpFixedSinCosPiINTEL:
+ case SPIRV::OpFixedRsqrtINTEL:
+ if (!ST.canUseExtension(
+ SPIRV::Extension::SPV_INTEL_arbitrary_precision_fixed_point))
+ report_fatal_error("This instruction requires the "
+ "following SPIR-V extension: "
+ "SPV_INTEL_arbitrary_precision_fixed_point",
+ false);
+ Reqs.addExtension(
+ SPIRV::Extension::SPV_INTEL_arbitrary_precision_fixed_point);
+ Reqs.addCapability(SPIRV::Capability::ArbitraryPrecisionFixedPointINTEL);
+ break;
case SPIRV::OpGroupIMulKHR:
case SPIRV::OpGroupFMulKHR:
case SPIRV::OpGroupBitwiseAndKHR:
diff --git a/llvm/lib/Target/SPIRV/SPIRVSymbolicOperands.td b/llvm/lib/Target/SPIRV/SPIRVSymbolicOperands.td
index 0db8a37f8683c..feb41d72b3d66 100644
--- a/llvm/lib/Target/SPIRV/SPIRVSymbolicOperands.td
+++ b/llvm/lib/Target/SPIRV/SPIRVSymbolicOperands.td
@@ -314,6 +314,7 @@ defm SPV_INTEL_long_composites : ExtensionOperand<117>;
defm SPV_INTEL_memory_access_aliasing : ExtensionOperand<118>;
defm SPV_INTEL_fp_max_error : ExtensionOperand<119>;
defm SPV_INTEL_ternary_bitwise_function : ExtensionOperand<120>;
+defm SPV_INTEL_arbitrary_precision_fixed_point : ExtensionOperand<121>;
//===----------------------------------------------------------------------===//
// Multiclass used to define Capabilities enum values and at the same time
@@ -514,6 +515,7 @@ defm LongCompositesINTEL : CapabilityOperand<6089, 0, 0, [SPV_INTEL_long_composi
defm BindlessImagesINTEL : CapabilityOperand<6528, 0, 0, [SPV_INTEL_bindless_images], []>;
defm MemoryAccessAliasingINTEL : CapabilityOperand<5910, 0, 0, [SPV_INTEL_memory_access_aliasing], []>;
defm FPMaxErrorINTEL : CapabilityOperand<6169, 0, 0, [SPV_INTEL_fp_max_error], []>;
+defm ArbitraryPrecisionFixedPointINTEL : CapabilityOperand<5922, 0, 0, [SPV_INTEL_arbitrary_precision_fixed_point], []>;
defm TernaryBitwiseFunctionINTEL : CapabilityOperand<6241, 0, 0, [SPV_INTEL_ternary_bitwise_function], []>;
//===----------------------------------------------------------------------===//
diff --git a/llvm/test/CodeGen/SPIRV/extensions/SPV_INTEL_arbitrary_precision_fixed_point/capability-arbitrary-precision-fixed-point-numbers.ll b/llvm/test/CodeGen/SPIRV/extensions/SPV_INTEL_arbitrary_precision_fixed_point/capability-arbitrary-precision-fixed-point-numbers.ll
new file mode 100644
index 0000000000000..bbab69378aab4
--- /dev/null
+++ b/llvm/test/CodeGen/SPIRV/extensions/SPV_INTEL_arbitrary_precision_fixed_point/capability-arbitrary-precision-fixed-point-numbers.ll
@@ -0,0 +1,563 @@
+; SYCL source (compiled with -S -emit-llvm -fsycl-device-only):
+; template <int W, int rW, bool S, int I, int rI>
+; void sqrt() {
+; ap_int<W> a;
+; auto ap_fixed_Sqrt = __spirv_FixedSqrtINTEL<W,rW>(a, S, I, rI);
+; ap_int<rW> b;
+; auto ap_fixed_Sqrt_b = __spirv_FixedSqrtINTEL<rW, W>(b, S, I, rI);
+; ap_int<rW> c;
+; auto ap_fixed_Sqrt_c = __spirv_FixedSqrtINTEL<rW, W>(c, S, I, rI);
+; }
+
+; template <int W, int rW, bool S, int I, int rI>
+; void recip() {
+; ap_int<W> a;
+; auto ap_fixed_Recip = __spirv_FixedRecipINTEL<W,rW>(a, S, I, rI);
+; }
+
+; template <int W, int rW, bool S, int I, int rI>
+; void rsqrt() {
+; ap_int<W> a;
+; auto ap_fixed_Rsqrt = __spirv_FixedRsqrtINTEL<W,rW>(a, S, I, rI);
+; }
+
+; template <int W, int rW, bool S, int I, int rI>
+; void sin() {
+; ap_int<W> a;
+; auto ap_fixed_Sin = __spirv_FixedSinINTEL<W,rW>(a, S, I, rI);
+; }
+
+; template <int W, int rW, bool S, int I, int rI>
+; void cos() {
+; ap_int<W> a;
+; auto ap_fixed_Cos = __spirv_FixedCosINTEL<W,rW>(a, S, I, rI);
+; }
+
+; template <int W, int rW, bool S, int I, int rI>
+; void sin_cos() {
+; ap_int<W> a;
+; auto ap_fixed_SinCos = __spirv_FixedSinCosINTEL<W,rW>(a, S, I, rI);
+; }
+
+; template <int W, int rW, bool S, int I, int rI>
+; void sin_pi() {
+; ap_int<W> a;
+; auto ap_fixed_SinPi = __spirv_FixedSinPiINTEL<W,rW>(a, S, I, rI);
+; }
+
+; template <int W, int rW, bool S, int I, int rI>
+; void cos_pi() {
+; ap_int<W> a;
+; auto ap_fixed_CosPi = __spirv_FixedCosPiINTEL<W,rW>(a, S, I, rI);
+; }
+
+; template <int W, int rW, bool S, int I, int rI>
+; void sin_cos_pi() {
+; ap_int<W> a;
+; auto ap_fixed_SinCosPi = __spirv_FixedSinCosPiINTEL<W,rW>(a, S, I, rI);
+; }
+
+; template <int W, int rW, bool S, int I, int rI>
+; void log() {
+; ap_int<W> a;
+; auto ap_fixed_Log = __spirv_FixedLogINTEL<W,rW>(a, S, I, rI);
+; }
+
+; template <int W, int rW, bool S, int I, int rI>
+; void exp() {
+; ap_int<W> a;
+; auto ap_fixed_Exp = __spirv_FixedExpINTEL<W,rW>(a, S, I, rI);
+; }
+
+; template <typename name, typename Func>
+; __attribute__((sycl_kernel)) void kernel_single_task(Func kernelFunc) {
+; kernelFunc();
+; }
+
+; int main() {
+; kernel_single_task<class kernel_function>([]() {
+; sqrt<13, 5, false, 2, 2>();
+; recip<3, 8, true, 4, 4>();
+; rsqrt<11, 10, false, 8, 6>();
+; sin<17, 11, true, 7, 5>();
+; cos<35, 28, false, 9, 3>();
+; sin_cos<31, 20, true, 10, 12>();
+; sin_pi<60, 5, false, 2, 2>();
+; cos_pi<28, 16, false, 8, 5>();
+; sin_cos_pi<13, 5, false, 2, 2>();
+; log<64, 44, true, 24, 22>();
+; exp<44, 34, false, 20, 20>();
+; exp<68, 68, false, 20, 20>();
+; });
+; return 0;
+; }
+
+; RUN: llc -verify-machineinstrs -O0 -mtriple=spirv64-unknown-unknown --spirv-ext=+SPV_INTEL_arbitrary_precision_fixed_point,+SPV_INTEL_arbitrary_precision_integers %s -o - | FileCheck %s
+; TODO: %if spirv-tools %{ llc -O0 -mtriple=spirv64-unknown-unknown --spirv-ext=+SPV_INTEL_arbitrary_precision_fixed_point,+SPV_INTEL_arbitrary_precision_integers %s -o - -filetype=obj | spirv-val %}
+
+
+; CHECK-DAG: OpCapability Kernel
+; CHECK-DAG: OpCapability ArbitraryPrecisionIntegersINTEL
+; CHECK-DAG: OpCapability ArbitraryPrecisionFixedPointINTEL
+; CHECK-DAG: OpExtension "SPV_INTEL_arbitrary_precision_fixed_point"
+; CHECK-DAG: OpExtension "SPV_INTEL_arbitrary_precision_integers"
+
+; CHECK-DAG: %[[Ty_8:[0-9]+]] = OpTypeInt 8 0
+; CHECK-DAG: %[[Ty_13:[0-9]+]] = OpTypeInt 13 0
+; CHECK-DAG: %[[Ty_5:[0-9]+]] = OpTypeInt 5 0
+; CHECK-DAG: %[[Ty_3:[0-9]+]] = OpTypeInt 3 0
+; CHECK-DAG: %[[Ty_11:[0-9]+]] = OpTypeInt 11 0
+; CHECK-DAG: %[[Ty_10:[0-9]+]] = OpTypeInt 10 0
+; CHECK-DAG: %[[Ty_17:[0-9]+]] = OpTypeInt 17 0
+; CHECK-DAG: %[[Ty_35:[0-9]+]] = OpTypeInt 35 0
+; CHECK-DAG: %[[Ty_28:[0-9]+]] = OpTypeInt 28 0
+; CHECK-DAG: %[[Ty_31:[0-9]+]] = OpTypeInt 31 0
+; CHECK-DAG: %[[Ty_40:[0-9]+]] = OpTypeInt 40 0
+; CHECK-DAG: %[[Ty_60:[0-9]+]] = OpTypeInt 60 0
+; CHECK-DAG: %[[Ty_16:[0-9]+]] = OpTypeInt 16 0
+; CHECK-DAG: %[[Ty_64:[0-9]+]] = OpTypeInt 64 0
+; CHECK-DAG: %[[Ty_44:[0-9]+]] = OpTypeInt 44 0
+; CHECK-DAG: %[[Ty_34:[0-9]+]] = OpTypeInt 34 0
+; CHECK-DAG: %[[Ty_51:[0-9]+]] = OpTypeInt 51 0
+
+
+
+; CHECK: %[[Sqrt_InId:[0-9]+]] = OpLoad %[[Ty_13]]
+; CHECK-NEXT: %[[#]] = OpFixedSqrtINTEL %[[Ty_5]] %[[Sqrt_InId]] 0 2 2 0 0
+
+; CHECK: %[[Sqrt_InId_B:[0-9]+]] = OpLoad %[[Ty_5]]
+; CHECK-NEXT: %[[#]] = OpFixedSqrtINTEL %[[Ty_13]] %[[Sqrt_InId_B]] 0 2 2 0 0
+
+; CHECK: %[[Sqrt_InId_C:[0-9]+]] = OpLoad %[[Ty_5]]
+; CHECK-NEXT: %[[#]] = OpFixedSqrtINTEL %[[Ty_13]] %[[Sqrt_InId_C]] 0 2 2 0 0
+
+
+; CHECK: %[[Recip_InId:[0-9]+]] = OpLoad %[[Ty_3]]
+; CHECK-NEXT: %[[#]] = OpFixedRecipINTEL %[[Ty_8]] %[[Recip_InId]] 1 4 4 0 0
+
+; CHECK: %[[Rsqrt_InId:[0-9]+]] = OpLoad %[[Ty_11]]
+; CHECK-NEXT: %[[#]] = OpFixedRsqrtINTEL %[[Ty_10]] %[[Rsqrt_InId]] 0 8 6 0 0
+
+; CHECK: %[[Sin_InId:[0-9]+]] = OpLoad %[[Ty_17]]
+; CHECK-NEXT: %[[#]] = OpFixedSinINTEL %[[Ty_11]] %[[Sin_InId]] 1 7 5 0 0
+
+; CHECK: %[[Cos_InId:[0-9]+]] = OpLoad %[[Ty_35]]
+; CHECK-NEXT: %[[#]] = OpFixedCosINTEL %[[Ty_28]] %[[Cos_InId]] 0 9 3 0 0
+
+; CHECK: %[[SinCos_InId:[0-9]+]] = OpLoad %[[Ty_31]]
+; CHECK-NEXT: %[[#]] = OpFixedSinCosINTEL %[[Ty_40]] %[[SinCos_InId]] 1 10 12 0 0
+
+; CHECK: %[[SinPi_InId:[0-9]+]] = OpLoad %[[Ty_60]]
+; CHECK-NEXT: %[[#]] = OpFixedSinPiINTEL %[[Ty_5]] %[[SinPi_InId]] 0 2 2 0 0
+
+; CHECK: %[[CosPi_InId:[0-9]+]] = OpLoad %[[Ty_28]]
+; CHECK-NEXT: %[[#]] = OpFixedCosPiINTEL %[[Ty_16]] %[[CosPi_InId]] 0 8 5 0 0
+
+; CHECK: %[[SinCosPi_InId:[0-9]+]] = OpLoad %[[Ty_13]]
+; CHECK-NEXT: %[[#]] = OpFixedSinCosPiINTEL %[[Ty_10]] %[[SinCosPi_InId]] 0 2 2 0 0
+
+; CHECK: %[[Log_InId:[0-9]+]] = OpLoad %[[Ty_64]]
+; CHECK-NEXT: %[[#]] = OpFixedLogINTEL %[[Ty_44]] %[[Log_InId]] 1 24 22 0 0
+
+; CHECK: %[[Exp_InId:[0-9]+]] = OpLoad %[[T...
[truncated]
|
5dc861e to
d691a1e
Compare
|
Ping |
MrSidims
reviewed
May 6, 2025
There was a problem hiding this comment.
Thank you for the contribution! There are few nits, other than that the patch is LGTM.
I have one logistical question though, @bashbaug I see this extensions (as well as few others) moved into Altera folder in SPIR-V Registry. Are there plans to rename the extensions themselves?
..._INTEL_arbitrary_precision_fixed_point/capability-arbitrary-precision-fixed-point-numbers.ll
Outdated
Show resolved
Hide resolved
..._INTEL_arbitrary_precision_fixed_point/capability-arbitrary-precision-fixed-point-numbers.ll
Outdated
Show resolved
Hide resolved
..._INTEL_arbitrary_precision_fixed_point/capability-arbitrary-precision-fixed-point-numbers.ll
Outdated
Show resolved
Hide resolved
…_point --Added test file for the extension SPV_INTEL_arbitrary_precision_fixed_point
MrSidims
approved these changes
May 7, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
--Added support for extension SPV_INTEL_arbitrary_precision_fixed_point
--Added test files for extension SPV_INTEL_arbitrary_precision_fixed_point