|
| 1 | +//===-- MathToROCDL.cpp - conversion from Math to rocdl calls -------------===// |
| 2 | +// |
| 3 | +// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
| 4 | +// See https://llvm.org/LICENSE.txt for license information. |
| 5 | +// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
| 6 | +// |
| 7 | +//===----------------------------------------------------------------------===// |
| 8 | + |
| 9 | +#include "mlir/Conversion/MathToROCDL/MathToROCDL.h" |
| 10 | +#include "mlir/Conversion/LLVMCommon/LoweringOptions.h" |
| 11 | +#include "mlir/Conversion/LLVMCommon/TypeConverter.h" |
| 12 | +#include "mlir/Dialect/Func/IR/FuncOps.h" |
| 13 | +#include "mlir/Dialect/LLVMIR/LLVMDialect.h" |
| 14 | +#include "mlir/Dialect/LLVMIR/ROCDLDialect.h" |
| 15 | +#include "mlir/Dialect/Math/IR/Math.h" |
| 16 | +#include "mlir/Dialect/Utils/IndexingUtils.h" |
| 17 | +#include "mlir/Dialect/Vector/IR/VectorOps.h" |
| 18 | +#include "mlir/IR/BuiltinDialect.h" |
| 19 | +#include "mlir/IR/PatternMatch.h" |
| 20 | +#include "mlir/Pass/Pass.h" |
| 21 | +#include "mlir/Transforms/DialectConversion.h" |
| 22 | + |
| 23 | +#include "../GPUCommon/GPUOpsLowering.h" |
| 24 | +#include "../GPUCommon/IndexIntrinsicsOpLowering.h" |
| 25 | +#include "../GPUCommon/OpToFuncCallLowering.h" |
| 26 | +#include "mlir/Conversion/GPUCommon/GPUCommonPass.h" |
| 27 | + |
| 28 | +namespace mlir { |
| 29 | +#define GEN_PASS_DEF_CONVERTMATHTOROCDL |
| 30 | +#include "mlir/Conversion/Passes.h.inc" |
| 31 | +} // namespace mlir |
| 32 | + |
| 33 | +using namespace mlir; |
| 34 | + |
| 35 | +#define DEBUG_TYPE "math-to-rocdl" |
| 36 | +#define DBGS() (llvm::dbgs() << "[" DEBUG_TYPE "]: ") |
| 37 | + |
| 38 | +template <typename OpTy> |
| 39 | +static void populateOpPatterns(LLVMTypeConverter &converter, |
| 40 | + RewritePatternSet &patterns, StringRef f32Func, |
| 41 | + StringRef f64Func) { |
| 42 | + patterns.add<ScalarizeVectorOpLowering<OpTy>>(converter); |
| 43 | + patterns.add<OpToFuncCallLowering<OpTy>>(converter, f32Func, f64Func); |
| 44 | +} |
| 45 | + |
| 46 | +void mlir::populateMathToROCDLConversionPatterns(LLVMTypeConverter &converter, |
| 47 | + RewritePatternSet &patterns) { |
| 48 | + // Handled by mathToLLVM: math::AbsIOp |
| 49 | + // Handled by mathToLLVM: math::CopySignOp |
| 50 | + // Handled by mathToLLVM: math::CountLeadingZerosOp |
| 51 | + // Handled by mathToLLVM: math::CountTrailingZerosOp |
| 52 | + // Handled by mathToLLVM: math::CgPopOp |
| 53 | + // Handled by mathToLLVM: math::FmaOp |
| 54 | + // FIXME: math::IPowIOp |
| 55 | + // FIXME: math::FPowIOp |
| 56 | + // Handled by mathToLLVM: math::RoundEvenOp |
| 57 | + // Handled by mathToLLVM: math::RoundOp |
| 58 | + // Handled by mathToLLVM: math::TruncOp |
| 59 | + populateOpPatterns<math::AbsFOp>(converter, patterns, "__ocml_fabs_f32", |
| 60 | + "__ocml_fabs_f64"); |
| 61 | + populateOpPatterns<math::AcosOp>(converter, patterns, "__ocml_acos_f32", |
| 62 | + "__ocml_acos_f64"); |
| 63 | + populateOpPatterns<math::AcoshOp>(converter, patterns, "__ocml_acosh_f32", |
| 64 | + "__ocml_acosh_f64"); |
| 65 | + populateOpPatterns<math::AsinOp>(converter, patterns, "__ocml_asin_f32", |
| 66 | + "__ocml_asin_f64"); |
| 67 | + populateOpPatterns<math::AsinhOp>(converter, patterns, "__ocml_asinh_f32", |
| 68 | + "__ocml_asinh_f64"); |
| 69 | + populateOpPatterns<math::AtanOp>(converter, patterns, "__ocml_atan_f32", |
| 70 | + "__ocml_atan_f64"); |
| 71 | + populateOpPatterns<math::AtanhOp>(converter, patterns, "__ocml_atanh_f32", |
| 72 | + "__ocml_atanh_f64"); |
| 73 | + populateOpPatterns<math::Atan2Op>(converter, patterns, "__ocml_atan2_f32", |
| 74 | + "__ocml_atan2_f64"); |
| 75 | + populateOpPatterns<math::CbrtOp>(converter, patterns, "__ocml_cbrt_f32", |
| 76 | + "__ocml_cbrt_f64"); |
| 77 | + populateOpPatterns<math::CeilOp>(converter, patterns, "__ocml_ceil_f32", |
| 78 | + "__ocml_ceil_f64"); |
| 79 | + populateOpPatterns<math::CosOp>(converter, patterns, "__ocml_cos_f32", |
| 80 | + "__ocml_cos_f64"); |
| 81 | + populateOpPatterns<math::CoshOp>(converter, patterns, "__ocml_cosh_f32", |
| 82 | + "__ocml_cosh_f64"); |
| 83 | + populateOpPatterns<math::SinhOp>(converter, patterns, "__ocml_sinh_f32", |
| 84 | + "__ocml_sinh_f64"); |
| 85 | + populateOpPatterns<math::ExpOp>(converter, patterns, "__ocml_exp_f32", |
| 86 | + "__ocml_exp_f64"); |
| 87 | + populateOpPatterns<math::Exp2Op>(converter, patterns, "__ocml_exp2_f32", |
| 88 | + "__ocml_exp2_f64"); |
| 89 | + populateOpPatterns<math::ExpM1Op>(converter, patterns, "__ocml_expm1_f32", |
| 90 | + "__ocml_expm1_f64"); |
| 91 | + populateOpPatterns<math::FloorOp>(converter, patterns, "__ocml_floor_f32", |
| 92 | + "__ocml_floor_f64"); |
| 93 | + populateOpPatterns<math::LogOp>(converter, patterns, "__ocml_log_f32", |
| 94 | + "__ocml_log_f64"); |
| 95 | + populateOpPatterns<math::Log10Op>(converter, patterns, "__ocml_log10_f32", |
| 96 | + "__ocml_log10_f64"); |
| 97 | + populateOpPatterns<math::Log1pOp>(converter, patterns, "__ocml_log1p_f32", |
| 98 | + "__ocml_log1p_f64"); |
| 99 | + populateOpPatterns<math::Log2Op>(converter, patterns, "__ocml_log2_f32", |
| 100 | + "__ocml_log2_f64"); |
| 101 | + populateOpPatterns<math::PowFOp>(converter, patterns, "__ocml_pow_f32", |
| 102 | + "__ocml_pow_f64"); |
| 103 | + populateOpPatterns<math::RsqrtOp>(converter, patterns, "__ocml_rsqrt_f32", |
| 104 | + "__ocml_rsqrt_f64"); |
| 105 | + populateOpPatterns<math::SinOp>(converter, patterns, "__ocml_sin_f32", |
| 106 | + "__ocml_sin_f64"); |
| 107 | + populateOpPatterns<math::SqrtOp>(converter, patterns, "__ocml_sqrt_f32", |
| 108 | + "__ocml_sqrt_f64"); |
| 109 | + populateOpPatterns<math::TanhOp>(converter, patterns, "__ocml_tanh_f32", |
| 110 | + "__ocml_tanh_f64"); |
| 111 | + populateOpPatterns<math::TanOp>(converter, patterns, "__ocml_tan_f32", |
| 112 | + "__ocml_tan_f64"); |
| 113 | + populateOpPatterns<math::ErfOp>(converter, patterns, "__ocml_erf_f32", |
| 114 | + "__ocml_erf_f64"); |
| 115 | + // Single arith pattern that needs a ROCDL call, probably not |
| 116 | + // worth creating a separate pass for it. |
| 117 | + populateOpPatterns<arith::RemFOp>(converter, patterns, "__ocml_fmod_f32", |
| 118 | + "__ocml_fmod_f64"); |
| 119 | +} |
| 120 | + |
| 121 | +namespace { |
| 122 | +struct ConvertMathToROCDLPass |
| 123 | + : public impl::ConvertMathToROCDLBase<ConvertMathToROCDLPass> { |
| 124 | + ConvertMathToROCDLPass() = default; |
| 125 | + void runOnOperation() override; |
| 126 | +}; |
| 127 | +} // namespace |
| 128 | + |
| 129 | +void ConvertMathToROCDLPass::runOnOperation() { |
| 130 | + auto m = getOperation(); |
| 131 | + MLIRContext *ctx = m.getContext(); |
| 132 | + |
| 133 | + RewritePatternSet patterns(&getContext()); |
| 134 | + LowerToLLVMOptions options(ctx, DataLayout(m)); |
| 135 | + LLVMTypeConverter converter(ctx, options); |
| 136 | + populateMathToROCDLConversionPatterns(converter, patterns); |
| 137 | + ConversionTarget target(getContext()); |
| 138 | + target.addLegalDialect<BuiltinDialect, func::FuncDialect, |
| 139 | + vector::VectorDialect, LLVM::LLVMDialect>(); |
| 140 | + target.addIllegalOp<LLVM::CosOp, LLVM::ExpOp, LLVM::Exp2Op, LLVM::FAbsOp, |
| 141 | + LLVM::FCeilOp, LLVM::FFloorOp, LLVM::FRemOp, LLVM::LogOp, |
| 142 | + LLVM::Log10Op, LLVM::Log2Op, LLVM::PowOp, LLVM::SinOp, |
| 143 | + LLVM::SqrtOp>(); |
| 144 | + if (failed(applyPartialConversion(m, target, std::move(patterns)))) |
| 145 | + signalPassFailure(); |
| 146 | +} |
0 commit comments