1 //===- QuantOps.cpp - Quantization Type and Ops Implementation --*- C++ -*-===// 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/Dialect/Quant/QuantOps.h" 10 #include "TypeDetail.h" 11 12 #include "mlir/Dialect/Quant/QuantTypes.h" 13 #include "mlir/IR/MLIRContext.h" 14 #include "mlir/IR/Matchers.h" 15 #include "mlir/IR/PatternMatch.h" 16 #include "mlir/IR/StandardTypes.h" 17 #include "llvm/ADT/StringRef.h" 18 #include "llvm/ADT/Twine.h" 19 #include "llvm/Support/MathExtras.h" 20 #include <numeric> 21 22 using namespace mlir; 23 using namespace mlir::quant; 24 using namespace mlir::quant::detail; 25 26 void QuantizationDialect::initialize() { 27 addTypes<AnyQuantizedType, UniformQuantizedType, 28 UniformQuantizedPerAxisType>(); 29 addOperations< 30 #define GET_OP_LIST 31 #include "mlir/Dialect/Quant/QuantOps.cpp.inc" 32 >(); 33 } 34 35 OpFoldResult StorageCastOp::fold(ArrayRef<Attribute> operands) { 36 // Matches x -> [scast -> scast] -> y, replacing the second scast with the 37 // value of x if the casts invert each other. 38 auto srcScastOp = arg().getDefiningOp<StorageCastOp>(); 39 if (!srcScastOp || srcScastOp.arg().getType() != getType()) 40 return OpFoldResult(); 41 return srcScastOp.arg(); 42 } 43 44 /// The quantization specification should match the expressed type. 45 static bool isValidQuantizationSpec(Attribute quantSpec, Type expressed) { 46 if (auto typeAttr = quantSpec.dyn_cast<TypeAttr>()) { 47 Type spec = typeAttr.getValue(); 48 if (spec.isa<TensorType, VectorType>()) 49 return false; 50 51 // The spec should be either a quantized type which is compatible to the 52 // expressed type, or a primitive type which is as same as the 53 // (element type of) the expressed type. 54 if (auto quantizedType = spec.dyn_cast<QuantizedType>()) 55 return quantizedType.isCompatibleExpressedType(expressed); 56 57 if (auto tensorType = expressed.dyn_cast<TensorType>()) 58 return spec == tensorType.getElementType(); 59 60 if (auto vectorType = expressed.dyn_cast<VectorType>()) 61 return spec == vectorType.getElementType(); 62 } 63 return false; 64 } 65 66 static LogicalResult verifyRegionOp(QuantizeRegionOp op) { 67 // There are specifications for both inputs and outputs. 68 if (op.getNumOperands() != op.input_specs().size() || 69 op.getNumResults() != op.output_specs().size()) 70 return op.emitOpError( 71 "has unmatched operands/results number and spec attributes number"); 72 73 // Verify that quantization specifications are valid. 74 for (auto input : llvm::zip(op.getOperandTypes(), op.input_specs())) { 75 Type inputType = std::get<0>(input); 76 Attribute inputSpec = std::get<1>(input); 77 if (!isValidQuantizationSpec(inputSpec, inputType)) { 78 return op.emitOpError() << "has incompatible specification " << inputSpec 79 << " and input type " << inputType; 80 } 81 } 82 83 for (auto result : llvm::zip(op.getResultTypes(), op.output_specs())) { 84 Type outputType = std::get<0>(result); 85 Attribute outputSpec = std::get<1>(result); 86 if (!isValidQuantizationSpec(outputSpec, outputType)) { 87 return op.emitOpError() << "has incompatible specification " << outputSpec 88 << " and output type " << outputType; 89 } 90 } 91 return success(); 92 } 93 94 #define GET_OP_CLASSES 95 #include "mlir/Dialect/Quant/QuantOps.cpp.inc" 96