[NFC] Change structure of conversion folder. (#96)

* Change structure of conversion folder.

* Fix comments.

Co-authored-by: Tian Jin <tjingrant@gmail.com>
This commit is contained in:
Gheorghe-Teodor Bercea 2020-02-25 10:38:08 -05:00 committed by GitHub
parent 32f08bcf0c
commit ee3e140ddb
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
15 changed files with 612 additions and 439 deletions

View File

@ -62,7 +62,21 @@ target_include_directories(onnf_shape_inference
target_link_libraries(onnf_shape_inference ${MLIRLibs})
add_dependencies(onnf_shape_inference gen_krnl_ops)
add_library(onnf_lower_frontend conversion/onnx_to_krnl/convert_onnx_to_krnl.cpp)
add_library(onnf_lower_frontend
conversion/onnx_to_krnl/onnx_to_krnl_common.cpp
conversion/onnx_to_krnl/onnx_to_krnl_common.hpp
conversion/onnx_to_krnl/math/elementwise.cpp
conversion/onnx_to_krnl/math/gemm.cpp
conversion/onnx_to_krnl/math/matmul.cpp
conversion/onnx_to_krnl/math/reduction.cpp
conversion/onnx_to_krnl/math/softmax.cpp
conversion/onnx_to_krnl/nn/conv.cpp
conversion/onnx_to_krnl/nn/normalization.cpp
conversion/onnx_to_krnl/tensor/identity.cpp
conversion/onnx_to_krnl/tensor/reshape.cpp
conversion/onnx_to_krnl/tensor/transpose.cpp
conversion/onnx_to_krnl/tensor/unsqueeze.cpp
conversion/onnx_to_krnl/convert_onnx_to_krnl.cpp)
target_include_directories(onnf_lower_frontend
PRIVATE ${ONNF_SRC_ROOT} ${ONNF_BIN_ROOT}
${ONNF_SRC_ROOT})

View File

@ -8,404 +8,11 @@
// Krnl IR and standard operations.
//
//===----------------------------------------------------------------------===//
#include <map>
#include "mlir/Dialect/AffineOps/AffineOps.h"
#include "mlir/Dialect/StandardOps/Ops.h"
#include "mlir/Pass/Pass.h"
#include "mlir/Transforms/DialectConversion.h"
#include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/Sequence.h"
#include "src/dialect/krnl/krnl_helper.hpp"
#include "src/dialect/krnl/krnl_ops.hpp"
#include "src/dialect/onnx/onnx_ops.hpp"
#include "src/pass/passes.hpp"
#include "src/conversion/onnx_to_krnl/onnx_to_krnl_common.hpp"
using namespace mlir;
//===----------------------------------------------------------------------===//
// FrontendToAffine RewritePatterns
//===----------------------------------------------------------------------===//
/// Check is all dimensions are known at compile time.
static bool hasAllConstantDimensions(MemRefType type) {
auto memRefShape = type.getShape();
for (int i = 0; i < memRefShape.size(); ++i)
if (memRefShape[i] < 0)
return false;
return true;
}
/// Get the corresponding MemRefType of a given TensorType/MemRefType.
static MemRefType convertToMemRefType(Type type) {
MemRefType memRefType;
auto tensorType = type.dyn_cast<TensorType>();
if (tensorType) {
assert(tensorType.hasRank() && "expected only ranked shapes");
memRefType =
MemRefType::get(tensorType.getShape(), tensorType.getElementType());
} else {
memRefType = type.dyn_cast<MemRefType>();
}
return memRefType;
}
/// Insert an allocation and deallocation for the given MemRefType.
static Value insertAllocAndDealloc(MemRefType type, Location loc,
PatternRewriter &rewriter,
bool insertDealloc,
ArrayRef<Value> operands = {}) {
// Put together alloc operands for any dynamic dimensions of the memref.
AllocOp alloc;
if (!operands.empty()) {
auto memRefShape = type.getShape();
auto rank = memRefShape.size();
std::map<int, Value> fromOperands;
for (int reversedIdx = 0; reversedIdx < rank; ++reversedIdx) {
int memRefDimIdx = rank - 1 - reversedIdx;
if (memRefShape[memRefDimIdx] < 0) { // unknown dimension
Value maxDim = nullptr;
for (int i = 0; i < operands.size(); i++) {
auto operandShape =
operands[i].getType().cast<MemRefType>().getShape();
int operandDimIdx = operandShape.size() - 1 - reversedIdx;
if (operandDimIdx < 0)
continue;
// In case of operations with broadcasting, the dimension of the
// alloc result is the maximum size along each dimension of the
// operands.
auto operandDim =
rewriter.create<DimOp>(loc, operands[i], operandDimIdx);
if (maxDim) {
auto maxCondition = rewriter.create<CmpIOp>(loc, CmpIPredicate::sgt,
operandDim, maxDim);
maxDim = rewriter.create<SelectOp>(loc, maxCondition, operandDim,
maxDim);
} else {
maxDim = operandDim;
}
}
fromOperands.insert(std::make_pair(memRefDimIdx, maxDim));
}
}
SmallVector<Value, 4> allocOperands;
for (int i = 0; i < rank; ++i)
if (memRefShape[i] < 0)
allocOperands.push_back(fromOperands[i]);
alloc = rewriter.create<AllocOp>(loc, type, allocOperands);
} else {
alloc = rewriter.create<AllocOp>(loc, type);
}
// Make sure to allocate at the beginning of the block if
// all dimensions are known.
auto *parentBlock = alloc.getOperation()->getBlock();
if (hasAllConstantDimensions(type))
alloc.getOperation()->moveBefore(&parentBlock->front());
if (insertDealloc) {
auto dealloc = rewriter.create<DeallocOp>(loc, alloc);
dealloc.getOperation()->moveBefore(&parentBlock->back());
}
return alloc;
}
// Determine if current function returns the result value of the
// current op being lowered. If it does then dealloc should not be
// inserted.
static bool checkInsertDealloc(Operation *currentOp) {
auto parentBlock = currentOp->getBlock();
bool insertDealloc = true;
parentBlock->walk([&insertDealloc, currentOp](ReturnOp op) {
assert(currentOp->getNumResults() < 2 &&
"No more than one result supported (for now).");
// If there is at least one result to investigate.
if (currentOp->getNumResults() > 0) {
auto result = currentOp->getResult(0);
for (const auto &operand : op.getOperands())
if (operand == result)
insertDealloc = false;
}
});
return insertDealloc;
}
// Create a mapping from result type's dimensions to input type's dimensions,
// given that the result type is the result of a reduction op over the input
// type.
std::map<int64_t, int64_t>
getReductionMapping(MemRefType inputTy, ArrayRef<int64_t> axes, bool keepdims) {
std::map<int64_t, int64_t> OutInDimMap;
int64_t rank = inputTy.getRank();
// Mark reduction axes.
std::vector<bool> isReductionAxis;
for (decltype(rank) i = 0; i < rank; ++i) {
if (std::find(axes.begin(), axes.end(), i) != axes.end())
isReductionAxis.push_back(true);
else
isReductionAxis.push_back(false);
}
for (decltype(rank) inIndex = 0, outIndex = 0; inIndex < rank; ++inIndex) {
// If it is a reduction axis, there is no relationship among dimensions.
if (isReductionAxis[inIndex]) {
if (keepdims)
outIndex++;
} else {
OutInDimMap.insert(std::make_pair(outIndex, inIndex));
outIndex++;
}
}
return OutInDimMap;
}
// Add bounds associated with the op operand to the KRNL iteration pack.
// Dynamic dimenions are supported.
static void addDimensionToPack(ConversionPatternRewriter &rewriter,
Location loc, KrnlIterateOperandPack &pack,
Value operand, int index) {
auto shape = operand.getType().cast<MemRefType>().getShape();
if (shape[index] < 0) {
pack.pushConstantBound(0);
pack.pushOperandBound(
rewriter.create<DimOp>(loc, operand, index).getResult());
} else {
pack.pushConstantBound(0);
pack.pushConstantBound(shape[index]);
}
}
// Function that defines the KRNL dialect loops and their respective
// optimized version.
static KrnlOptimizeLoopsOp
emitOptimizedLoops(ConversionPatternRewriter &rewriter, Location loc,
std::vector<Value> &loops,
std::vector<Value> &optimizedLoops, int64_t numLoops) {
// Define loops.
auto loopsOp = rewriter.create<KrnlDefineLoopsOp>(loc, numLoops);
loops.reserve(numLoops);
for (auto result : loopsOp.getResults())
loops.push_back(result);
// Define optimized version of the loops.
auto optimizedLoopsOp = rewriter.create<KrnlOptimizeLoopsOp>(loc, numLoops);
optimizedLoops.reserve(numLoops);
for (auto result : optimizedLoopsOp.getResults())
optimizedLoops.push_back(result);
return optimizedLoopsOp;
}
// Function that emits the loops and their optimized version.
// The function returns a reference to the inner optimization block.
static Block *defineLoops(ConversionPatternRewriter &rewriter, Location loc,
std::vector<Value> &loops,
std::vector<Value> &optimizedLoops,
int64_t numLoops) {
KrnlOptimizeLoopsOp optimizedLoopsOp =
emitOptimizedLoops(rewriter, loc, loops, optimizedLoops, numLoops);
return &optimizedLoopsOp.region().front();
}
// Function which emits a basic set of loops and optimized loops
// for a given operation argument. A reference to the loop optimization
// block is returned in the last argument of the function.
static void emitKrnlLoopsAndIterationForOperand(
ConversionPatternRewriter &rewriter, Location loc, Value operand,
std::vector<Value> &originalLoops, KrnlOptimizeLoopsOp &optimizedLoopsOp,
KrnlIterateOp &iterateOp) {
// Operand shape.
auto shape = operand.getType().cast<MemRefType>().getShape();
// Number of loops.
int64_t rank = shape.size();
// Define loops and optimized loops.
std::vector<Value> optimizedLoops;
optimizedLoopsOp =
emitOptimizedLoops(rewriter, loc, originalLoops, optimizedLoops, rank);
KrnlIterateOperandPack pack(rewriter, originalLoops, optimizedLoops);
// Iterate over the loop nest.
for (int i = 0; i < rank; ++i)
addDimensionToPack(rewriter, loc, pack, operand, i);
iterateOp = rewriter.create<KrnlIterateOp>(loc, pack);
}
unsigned getMemRefEltSizeInBytes(MemRefType memRefType) {
auto elementType = memRefType.getElementType();
unsigned sizeInBits;
if (elementType.isIntOrFloat()) {
sizeInBits = elementType.getIntOrFloatBitWidth();
} else {
auto vectorType = elementType.cast<VectorType>();
sizeInBits =
vectorType.getElementTypeBitWidth() * vectorType.getNumElements();
}
return llvm::divideCeil(sizeInBits, 8);
}
// Get run-time dimension information for unknown dimensions used for
// broadcasting.
std::map<int, std::map<int, Value>>
getBroadcastedDimInfo(Location loc, ConversionPatternRewriter &rewriter,
MemRefType memRefType, ArrayRef<Value> operands) {
auto memRefShape = memRefType.getShape();
int64_t rank = memRefShape.size();
// For unknown dimensions, we need to get dimension values at runtime in
// order to do broadcasting.
std::map<int, std::map<int, Value>> DimInfo;
// For each result dimension, compute the number of sharing operands.
// Sharing operands are operands sharing the same index (counting from the
// rightmost to the leftmost) for a given dimension.
std::map<int, int> sharedDimCount;
for (int reversedIdx = 0; reversedIdx < rank; ++reversedIdx) {
int dimIdx = rank - 1 - reversedIdx;
sharedDimCount[dimIdx] = 0;
for (int i = 0; i < operands.size(); ++i) {
auto shape = operands[i].getType().cast<MemRefType>().getShape();
if (reversedIdx <= shape.size() - 1)
sharedDimCount[dimIdx]++;
}
}
// An unknown dimension can have a value of 1 or N (N > 1).
// If its value is 1, it is broadcasted dimension.
// Otherwise, non-broadcasted dimension.
// We only care about unknown dimensions whose number of sharing operands is
// more than one, since they are potentially broadcasted dimensions.
for (int i = 0; i < operands.size(); ++i) {
std::map<int, Value> broadcastedDims;
auto shape = operands[i].getType().cast<MemRefType>().getShape();
int size = shape.size();
for (int j = 0; j < shape.size(); ++j) {
if (shape[j] < 0 and sharedDimCount[rank - size + j] > 1) {
auto dim = rewriter.create<DimOp>(loc, operands[i], j).getResult();
auto one = rewriter.create<ConstantIndexOp>(loc, 1);
auto isBroadcasted =
rewriter.create<CmpIOp>(loc, CmpIPredicate::eq, dim, one);
broadcastedDims.insert(std::make_pair(j, isBroadcasted));
}
}
DimInfo.insert(std::make_pair(i, broadcastedDims));
}
return DimInfo;
}
// Extract induction variables that are used for broadcasting values of a
// given operand.
std::vector<Value>
getLoopIVsForBroadcasting(Location loc, ConversionPatternRewriter &rewriter,
ArrayRef<Value> loopIVs, Value operand,
std::map<int, Value> broadcastedDims) {
// `operand` must has a ranked type. This should have been checked by the
// shape inference pass.
auto operandShape = operand.getType().cast<MemRefType>().getShape();
auto rank = operandShape.size();
auto loopCount = loopIVs.size();
std::vector<Value> newLoopIVs;
for (unsigned reversedIdx = 0; reversedIdx < rank; ++reversedIdx) {
auto dimIdx = rank - 1 - reversedIdx;
auto loopIdx = loopCount - 1 - reversedIdx;
if (operandShape[dimIdx] == 1) {
// Broadcasted dimension
auto zero = rewriter.create<ConstantIndexOp>(loc, 0);
newLoopIVs.insert(newLoopIVs.begin(), zero);
} else if ((operandShape[dimIdx] == -1) &&
(broadcastedDims.find(dimIdx) != broadcastedDims.end())) {
// Unknown dimension, it can have a value of 1 or N (N > 1).
// If its value is 1, it is broadcasted dimension.
// Otherwise, non-broadcasted dimension.
auto zero = rewriter.create<ConstantIndexOp>(loc, 0);
auto idx = rewriter.create<SelectOp>(loc, broadcastedDims[dimIdx], zero,
loopIVs[loopIdx]);
newLoopIVs.insert(newLoopIVs.begin(), idx);
} else {
// Non-broadcasted dimension
newLoopIVs.insert(newLoopIVs.begin(), loopIVs[loopIdx]);
}
}
return newLoopIVs;
}
namespace {
// This is to get a scalar operation of a given type for a specific operation.
template <typename Op>
struct ScalarOp {
using FOp = void;
using IOp = void;
};
template <typename FOp>
using ScalarFOp = typename ScalarOp<FOp>::FOp;
template <typename IOp>
using ScalarIOp = typename ScalarOp<IOp>::IOp;
// Get the identity element of a operation.
// Return NULL if the function does not have identity.
template <typename DataType, typename Op>
DataType getIdentityValue() {
return NULL;
}
//===----------------------------------------------------------------------===//
// This is used in the innermost loop of a KrnlIterateOp to insert computation
// composed of one or many scalar ops.
// Use template specialization for each of different ONNX operations.
//===----------------------------------------------------------------------===//
template <typename Op>
Value mapToLowerScalarOp(Operation *op, ArrayRef<Type> result_types,
ArrayRef<Value> operands,
ConversionPatternRewriter &rewriter) {
auto loc = op->getLoc();
Type element_type = operands.front().getType();
if (element_type.isa<IntegerType>()) {
return rewriter.create<ScalarIOp<Op>>(loc, result_types, operands,
mlir::None);
} else if (element_type.isa<FloatType>()) {
return rewriter.create<ScalarFOp<Op>>(loc, result_types, operands,
mlir::None);
} else {
emitError(loc, "unsupported element type");
return nullptr;
}
}
// We divide the operator lowering into different categories.
// These categories are mostly similar to the operator categories in ONNX:
// https://github.com/onnx/onnx/tree/master/onnx/defs.
// Besides, it is better to put operators with the same computation pattern into
// the same category, e.g. element-wise operators will belong to the elementwise
// category.
// Math
#include "src/conversion/onnx_to_krnl/rewrite_patterns/math/elementwise.inc"
#include "src/conversion/onnx_to_krnl/rewrite_patterns/math/gemm.inc"
#include "src/conversion/onnx_to_krnl/rewrite_patterns/math/reduction.inc"
#include "src/conversion/onnx_to_krnl/rewrite_patterns/math/softmax.inc"
#include "src/conversion/onnx_to_krnl/rewrite_patterns/math/matmul.inc"
// Tensor
#include "src/conversion/onnx_to_krnl/rewrite_patterns/tensor/identity.inc"
#include "src/conversion/onnx_to_krnl/rewrite_patterns/tensor/reshape.inc"
#include "src/conversion/onnx_to_krnl/rewrite_patterns/tensor/transpose.inc"
#include "src/conversion/onnx_to_krnl/rewrite_patterns/tensor/unsqueeze.inc"
// Neural network
#include "src/conversion/onnx_to_krnl/rewrite_patterns/nn/conv.inc"
#include "src/conversion/onnx_to_krnl/rewrite_patterns/nn/normalization.inc"
//===----------------------------------------------------------------------===//
// EntryPoint Op lowering to Krnl Entry Point.
//===----------------------------------------------------------------------===//
@ -427,39 +34,6 @@ public:
}
};
//===----------------------------------------------------------------------===//
// Conversion from Tensor type to the Standard dialect MemRef type.
//===----------------------------------------------------------------------===//
struct TensorTypeConverter : public TypeConverter {
using TypeConverter::TypeConverter;
TensorTypeConverter() {
addConversion(convertType);
}
static LogicalResult convertType(Type t, SmallVectorImpl<Type> &results) {
if (auto type = convertToMemRefType(t)) {
results.push_back(type);
return success();
}
results.push_back(t);
return success();
}
/// Return true if the inputs and outputs of the given function type are
/// legal. [Taken from MLIR and adapted to only check the legality of the
/// inputs. Once unranked results can be handled gracefully this
/// override needs to be removed in favour of the original MLIR one.]
bool isSignatureLegal(FunctionType funcType) {
return llvm::all_of(funcType.getInputs(),
[this](Type type) { return isLegal(type); });
}
};
} // end anonymous namespace.
//===----------------------------------------------------------------------===//
// Frontend to Krnl Dialect lowering pass
//===----------------------------------------------------------------------===//

View File

@ -1,4 +1,4 @@
//===----- elementwise.inc - Elementwise Ops ------------------------------===//
//===----- elementwise.cpp - Elementwise Ops ------------------------------===//
//
// Copyright 2019 The IBM Research Authors.
//
@ -8,6 +8,10 @@
//
//===----------------------------------------------------------------------===//
#include "src/conversion/onnx_to_krnl/onnx_to_krnl_common.hpp"
using namespace mlir;
template <>
struct ScalarOp<ONNXAddOp> {
using FOp = AddFOp;

View File

@ -1,4 +1,4 @@
//===----- gemm.inc - Lowering Gemm Op ------------------------------------===//
//===----- gemm.cpp - Lowering Gemm Op ------------------------------------===//
//
// Copyright 2019 The IBM Research Authors.
//
@ -8,6 +8,10 @@
//
//===----------------------------------------------------------------------===//
#include "src/conversion/onnx_to_krnl/onnx_to_krnl_common.hpp"
using namespace mlir;
template <typename GemmOp>
struct ONNXGemmOpLowering : public ConversionPattern {
ONNXGemmOpLowering(MLIRContext *ctx)

View File

@ -1,4 +1,4 @@
//===----- matmul.inc - Lowering Matmul Op --------------------------------===//
//===----- matmul.cpp - Lowering Matmul Op --------------------------------===//
//
// Copyright 2019 The IBM Research Authors.
//
@ -8,6 +8,10 @@
//
//===----------------------------------------------------------------------===//
#include "src/conversion/onnx_to_krnl/onnx_to_krnl_common.hpp"
using namespace mlir;
struct ONNXMatMulOpLowering : public ConversionPattern {
ONNXMatMulOpLowering(MLIRContext *ctx)
: ConversionPattern(mlir::ONNXMatMulOp::getOperationName(), 1, ctx) {}

View File

@ -1,4 +1,4 @@
//===----- reduction.inc - Lowering Reduction Ops -------------------------===//
//===----- reduction.cpp - Lowering Reduction Ops -------------------------===//
//
// Copyright 2019 The IBM Research Authors.
//
@ -8,6 +8,10 @@
//
//===----------------------------------------------------------------------===//
#include "src/conversion/onnx_to_krnl/onnx_to_krnl_common.hpp"
using namespace mlir;
// Identity values
template <>
float getIdentityValue<float, ONNXReduceMaxOp>(){

View File

@ -1,4 +1,4 @@
//===----- softmax.inc - Softmax Op ---------------------------------------===//
//===----- softmax.cpp - Softmax Op ---------------------------------------===//
//
// Copyright 2019 The IBM Research Authors.
//
@ -8,6 +8,10 @@
//
//===----------------------------------------------------------------------===//
#include "src/conversion/onnx_to_krnl/onnx_to_krnl_common.hpp"
using namespace mlir;
struct ONNXSoftmaxOpLowering : public ConversionPattern {
ONNXSoftmaxOpLowering(MLIRContext *ctx)
: ConversionPattern(mlir::ONNXSoftmaxOp::getOperationName(), 1, ctx) {}

View File

@ -1,4 +1,4 @@
//===----- conv.inc - Lowering Convolution Op -----------------------------===//
//===----- conv.cpp - Lowering Convolution Op -----------------------------===//
//
// Copyright 2019 The IBM Research Authors.
//
@ -8,6 +8,10 @@
//
//===----------------------------------------------------------------------===//
#include "src/conversion/onnx_to_krnl/onnx_to_krnl_common.hpp"
using namespace mlir;
struct ONNXConvNoBiasOpLowering : public ConversionPattern {
ONNXConvNoBiasOpLowering(MLIRContext *ctx)
: ConversionPattern(mlir::ONNXConvNoBiasOp::getOperationName(), 1, ctx) {}

View File

@ -1,4 +1,4 @@
//===----- normalization.inc - Lowering Normalization Ops -----------------===//
//===----- normalization.cpp - Lowering Normalization Ops -----------------===//
//
// Copyright 2019 The IBM Research Authors.
//
@ -8,6 +8,10 @@
//
//===----------------------------------------------------------------------===//
#include "src/conversion/onnx_to_krnl/onnx_to_krnl_common.hpp"
using namespace mlir;
struct ONNXBatchNormalizationTestModeOpLowering : public ConversionPattern {
ONNXBatchNormalizationTestModeOpLowering(MLIRContext *ctx)
: ConversionPattern(

View File

@ -0,0 +1,324 @@
//====-- onnx_to_krnl_common.cpp - ONNX dialects to Krnl lowering ---------===//
//
// Copyright 2019 The IBM Research Authors.
//
// =============================================================================
//
// This file contains common code shared by the functions performing the
// lowering to the KRNL dialect.
//
//===----------------------------------------------------------------------===//
#include "src/conversion/onnx_to_krnl/onnx_to_krnl_common.hpp"
/// Check is all dimensions are known at compile time.
bool hasAllConstantDimensions(MemRefType type) {
auto memRefShape = type.getShape();
for (int i = 0; i < memRefShape.size(); ++i)
if (memRefShape[i] < 0)
return false;
return true;
}
/// Get the corresponding MemRefType of a given TensorType/MemRefType.
MemRefType convertToMemRefType(Type type) {
MemRefType memRefType;
auto tensorType = type.dyn_cast<TensorType>();
if (tensorType) {
assert(tensorType.hasRank() && "expected only ranked shapes");
memRefType =
MemRefType::get(tensorType.getShape(), tensorType.getElementType());
} else {
memRefType = type.dyn_cast<MemRefType>();
}
return memRefType;
}
/// Insert an allocation and deallocation for the given MemRefType.
Value insertAllocAndDealloc(MemRefType type, Location loc,
PatternRewriter &rewriter,
bool insertDealloc,
ArrayRef<Value> operands) {
// Put together alloc operands for any dynamic dimensions of the memref.
AllocOp alloc;
if (!operands.empty()) {
auto memRefShape = type.getShape();
auto rank = memRefShape.size();
std::map<int, Value> fromOperands;
for (int reversedIdx = 0; reversedIdx < rank; ++reversedIdx) {
int memRefDimIdx = rank - 1 - reversedIdx;
if (memRefShape[memRefDimIdx] < 0) { // unknown dimension
Value maxDim = nullptr;
for (int i = 0; i < operands.size(); i++) {
auto operandShape =
operands[i].getType().cast<MemRefType>().getShape();
int operandDimIdx = operandShape.size() - 1 - reversedIdx;
if (operandDimIdx < 0)
continue;
// In case of operations with broadcasting, the dimension of the
// alloc result is the maximum size along each dimension of the
// operands.
auto operandDim =
rewriter.create<DimOp>(loc, operands[i], operandDimIdx);
if (maxDim) {
auto maxCondition = rewriter.create<CmpIOp>(loc, CmpIPredicate::sgt,
operandDim, maxDim);
maxDim = rewriter.create<SelectOp>(loc, maxCondition, operandDim,
maxDim);
} else {
maxDim = operandDim;
}
}
fromOperands.insert(std::make_pair(memRefDimIdx, maxDim));
}
}
SmallVector<Value, 4> allocOperands;
for (int i = 0; i < rank; ++i)
if (memRefShape[i] < 0)
allocOperands.push_back(fromOperands[i]);
alloc = rewriter.create<AllocOp>(loc, type, allocOperands);
} else {
alloc = rewriter.create<AllocOp>(loc, type);
}
// Make sure to allocate at the beginning of the block if
// all dimensions are known.
auto *parentBlock = alloc.getOperation()->getBlock();
if (hasAllConstantDimensions(type))
alloc.getOperation()->moveBefore(&parentBlock->front());
if (insertDealloc) {
auto dealloc = rewriter.create<DeallocOp>(loc, alloc);
dealloc.getOperation()->moveBefore(&parentBlock->back());
}
return alloc;
}
// Determine if current function returns the result value of the
// current op being lowered. If it does then dealloc should not be
// inserted.
bool checkInsertDealloc(Operation *currentOp) {
auto parentBlock = currentOp->getBlock();
bool insertDealloc = true;
parentBlock->walk([&insertDealloc, currentOp](ReturnOp op) {
assert(currentOp->getNumResults() < 2 &&
"No more than one result supported (for now).");
// If there is at least one result to investigate.
if (currentOp->getNumResults() > 0) {
auto result = currentOp->getResult(0);
for (const auto &operand : op.getOperands())
if (operand == result)
insertDealloc = false;
}
});
return insertDealloc;
}
// Create a mapping from result type's dimensions to input type's dimensions,
// given that the result type is the result of a reduction op over the input
// type.
std::map<int64_t, int64_t>
getReductionMapping(MemRefType inputTy, ArrayRef<int64_t> axes, bool keepdims) {
std::map<int64_t, int64_t> OutInDimMap;
int64_t rank = inputTy.getRank();
// Mark reduction axes.
std::vector<bool> isReductionAxis;
for (decltype(rank) i = 0; i < rank; ++i) {
if (std::find(axes.begin(), axes.end(), i) != axes.end())
isReductionAxis.push_back(true);
else
isReductionAxis.push_back(false);
}
for (decltype(rank) inIndex = 0, outIndex = 0; inIndex < rank; ++inIndex) {
// If it is a reduction axis, there is no relationship among dimensions.
if (isReductionAxis[inIndex]) {
if (keepdims)
outIndex++;
} else {
OutInDimMap.insert(std::make_pair(outIndex, inIndex));
outIndex++;
}
}
return OutInDimMap;
}
// Add bounds associated with the op operand to the KRNL iteration pack.
// Dynamic dimenions are supported.
void addDimensionToPack(ConversionPatternRewriter &rewriter,
Location loc, KrnlIterateOperandPack &pack,
Value operand, int index) {
auto shape = operand.getType().cast<MemRefType>().getShape();
if (shape[index] < 0) {
pack.pushConstantBound(0);
pack.pushOperandBound(
rewriter.create<DimOp>(loc, operand, index).getResult());
} else {
pack.pushConstantBound(0);
pack.pushConstantBound(shape[index]);
}
}
// Function that defines the KRNL dialect loops and their respective
// optimized version.
KrnlOptimizeLoopsOp
emitOptimizedLoops(ConversionPatternRewriter &rewriter, Location loc,
std::vector<Value> &loops,
std::vector<Value> &optimizedLoops, int64_t numLoops) {
// Define loops.
auto loopsOp = rewriter.create<KrnlDefineLoopsOp>(loc, numLoops);
loops.reserve(numLoops);
for (auto result : loopsOp.getResults())
loops.push_back(result);
// Define optimized version of the loops.
auto optimizedLoopsOp = rewriter.create<KrnlOptimizeLoopsOp>(loc, numLoops);
optimizedLoops.reserve(numLoops);
for (auto result : optimizedLoopsOp.getResults())
optimizedLoops.push_back(result);
return optimizedLoopsOp;
}
// Function that emits the loops and their optimized version.
// The function returns a reference to the inner optimization block.
Block *defineLoops(ConversionPatternRewriter &rewriter, Location loc,
std::vector<Value> &loops,
std::vector<Value> &optimizedLoops,
int64_t numLoops) {
KrnlOptimizeLoopsOp optimizedLoopsOp =
emitOptimizedLoops(rewriter, loc, loops, optimizedLoops, numLoops);
return &optimizedLoopsOp.region().front();
}
// Function which emits a basic set of loops and optimized loops
// for a given operation argument. A reference to the loop optimization
// block is returned in the last argument of the function.
void emitKrnlLoopsAndIterationForOperand(
ConversionPatternRewriter &rewriter, Location loc, Value operand,
std::vector<Value> &originalLoops, KrnlOptimizeLoopsOp &optimizedLoopsOp,
KrnlIterateOp &iterateOp) {
// Operand shape.
auto shape = operand.getType().cast<MemRefType>().getShape();
// Number of loops.
int64_t rank = shape.size();
// Define loops and optimized loops.
std::vector<Value> optimizedLoops;
optimizedLoopsOp =
emitOptimizedLoops(rewriter, loc, originalLoops, optimizedLoops, rank);
KrnlIterateOperandPack pack(rewriter, originalLoops, optimizedLoops);
// Iterate over the loop nest.
for (int i = 0; i < rank; ++i)
addDimensionToPack(rewriter, loc, pack, operand, i);
iterateOp = rewriter.create<KrnlIterateOp>(loc, pack);
}
unsigned getMemRefEltSizeInBytes(MemRefType memRefType) {
auto elementType = memRefType.getElementType();
unsigned sizeInBits;
if (elementType.isIntOrFloat()) {
sizeInBits = elementType.getIntOrFloatBitWidth();
} else {
auto vectorType = elementType.cast<VectorType>();
sizeInBits =
vectorType.getElementTypeBitWidth() * vectorType.getNumElements();
}
return llvm::divideCeil(sizeInBits, 8);
}
// Get run-time dimension information for unknown dimensions used for
// broadcasting.
std::map<int, std::map<int, Value>>
getBroadcastedDimInfo(Location loc, ConversionPatternRewriter &rewriter,
MemRefType memRefType, ArrayRef<Value> operands) {
auto memRefShape = memRefType.getShape();
int64_t rank = memRefShape.size();
// For unknown dimensions, we need to get dimension values at runtime in
// order to do broadcasting.
std::map<int, std::map<int, Value>> DimInfo;
// For each result dimension, compute the number of sharing operands.
// Sharing operands are operands sharing the same index (counting from the
// rightmost to the leftmost) for a given dimension.
std::map<int, int> sharedDimCount;
for (int reversedIdx = 0; reversedIdx < rank; ++reversedIdx) {
int dimIdx = rank - 1 - reversedIdx;
sharedDimCount[dimIdx] = 0;
for (int i = 0; i < operands.size(); ++i) {
auto shape = operands[i].getType().cast<MemRefType>().getShape();
if (reversedIdx <= shape.size() - 1)
sharedDimCount[dimIdx]++;
}
}
// An unknown dimension can have a value of 1 or N (N > 1).
// If its value is 1, it is broadcasted dimension.
// Otherwise, non-broadcasted dimension.
// We only care about unknown dimensions whose number of sharing operands is
// more than one, since they are potentially broadcasted dimensions.
for (int i = 0; i < operands.size(); ++i) {
std::map<int, Value> broadcastedDims;
auto shape = operands[i].getType().cast<MemRefType>().getShape();
int size = shape.size();
for (int j = 0; j < shape.size(); ++j) {
if (shape[j] < 0 and sharedDimCount[rank - size + j] > 1) {
auto dim = rewriter.create<DimOp>(loc, operands[i], j).getResult();
auto one = rewriter.create<ConstantIndexOp>(loc, 1);
auto isBroadcasted =
rewriter.create<CmpIOp>(loc, CmpIPredicate::eq, dim, one);
broadcastedDims.insert(std::make_pair(j, isBroadcasted));
}
}
DimInfo.insert(std::make_pair(i, broadcastedDims));
}
return DimInfo;
}
// Extract induction variables that are used for broadcasting values of a
// given operand.
std::vector<Value>
getLoopIVsForBroadcasting(Location loc, ConversionPatternRewriter &rewriter,
ArrayRef<Value> loopIVs, Value operand,
std::map<int, Value> broadcastedDims) {
// `operand` must has a ranked type. This should have been checked by the
// shape inference pass.
auto operandShape = operand.getType().cast<MemRefType>().getShape();
auto rank = operandShape.size();
auto loopCount = loopIVs.size();
std::vector<Value> newLoopIVs;
for (unsigned reversedIdx = 0; reversedIdx < rank; ++reversedIdx) {
auto dimIdx = rank - 1 - reversedIdx;
auto loopIdx = loopCount - 1 - reversedIdx;
if (operandShape[dimIdx] == 1) {
// Broadcasted dimension
auto zero = rewriter.create<ConstantIndexOp>(loc, 0);
newLoopIVs.insert(newLoopIVs.begin(), zero);
} else if ((operandShape[dimIdx] == -1) &&
(broadcastedDims.find(dimIdx) != broadcastedDims.end())) {
// Unknown dimension, it can have a value of 1 or N (N > 1).
// If its value is 1, it is broadcasted dimension.
// Otherwise, non-broadcasted dimension.
auto zero = rewriter.create<ConstantIndexOp>(loc, 0);
auto idx = rewriter.create<SelectOp>(loc, broadcastedDims[dimIdx], zero,
loopIVs[loopIdx]);
newLoopIVs.insert(newLoopIVs.begin(), idx);
} else {
// Non-broadcasted dimension
newLoopIVs.insert(newLoopIVs.begin(), loopIVs[loopIdx]);
}
}
return newLoopIVs;
}

View File

@ -0,0 +1,217 @@
//====-- onnx_to_krnl_common.hpp - ONNX dialects to Krnl lowering ---------===//
//
// Copyright 2019 The IBM Research Authors.
//
// =============================================================================
//
// This file contains common code shared by the functions performing the
// lowering to the KRNL dialect.
//
//===----------------------------------------------------------------------===//
#pragma once
#include <map>
#include "mlir/Dialect/AffineOps/AffineOps.h"
#include "mlir/Dialect/StandardOps/Ops.h"
#include "mlir/Pass/Pass.h"
#include "mlir/Transforms/DialectConversion.h"
#include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/Sequence.h"
#include "mlir/IR/PatternMatch.h"
#include "src/dialect/krnl/krnl_helper.hpp"
#include "src/dialect/krnl/krnl_ops.hpp"
#include "src/dialect/onnx/onnx_ops.hpp"
#include "src/pass/passes.hpp"
using namespace mlir;
//===----------------------------------------------------------------------===//
// Common functions used when lowering the ONNX frontend dialect to KRNL.
//===----------------------------------------------------------------------===//
/// Check is all dimensions are known at compile time.
bool hasAllConstantDimensions(MemRefType type);
/// Get the corresponding MemRefType of a given TensorType/MemRefType.
MemRefType convertToMemRefType(Type type);
/// Insert an allocation and deallocation for the given MemRefType.
Value insertAllocAndDealloc(MemRefType type, Location loc,
PatternRewriter &rewriter,
bool insertDealloc,
ArrayRef<Value> operands = {});
// Determine if current function returns the result value of the
// current op being lowered. If it does then dealloc should not be
// inserted.
bool checkInsertDealloc(Operation *currentOp);
// Create a mapping from result type's dimensions to input type's dimensions,
// given that the result type is the result of a reduction op over the input
// type.
std::map<int64_t, int64_t>
getReductionMapping(MemRefType inputTy, ArrayRef<int64_t> axes, bool keepdims);
// Add bounds associated with the op operand to the KRNL iteration pack.
// Dynamic dimenions are supported.
void addDimensionToPack(ConversionPatternRewriter &rewriter,
Location loc, KrnlIterateOperandPack &pack,
Value operand, int index);
// Function that defines the KRNL dialect loops and their respective
// optimized version.
KrnlOptimizeLoopsOp
emitOptimizedLoops(ConversionPatternRewriter &rewriter, Location loc,
std::vector<Value> &loops,
std::vector<Value> &optimizedLoops, int64_t numLoops);
// Function that emits the loops and their optimized version.
// The function returns a reference to the inner optimization block.
Block *defineLoops(ConversionPatternRewriter &rewriter, Location loc,
std::vector<Value> &loops,
std::vector<Value> &optimizedLoops,
int64_t numLoops);
// Function which emits a basic set of loops and optimized loops
// for a given operation argument. A reference to the loop optimization
// block is returned in the last argument of the function.
void emitKrnlLoopsAndIterationForOperand(
ConversionPatternRewriter &rewriter, Location loc, Value operand,
std::vector<Value> &originalLoops, KrnlOptimizeLoopsOp &optimizedLoopsOp,
KrnlIterateOp &iterateOp);
unsigned getMemRefEltSizeInBytes(MemRefType memRefType);
// Get run-time dimension information for unknown dimensions used for
// broadcasting.
std::map<int, std::map<int, Value>>
getBroadcastedDimInfo(Location loc, ConversionPatternRewriter &rewriter,
MemRefType memRefType, ArrayRef<Value> operands);
// Extract induction variables that are used for broadcasting values of a
// given operand.
std::vector<Value>
getLoopIVsForBroadcasting(Location loc, ConversionPatternRewriter &rewriter,
ArrayRef<Value> loopIVs, Value operand,
std::map<int, Value> broadcastedDims);
//===----------------------------------------------------------------------===//
// This is to get a scalar operation of a given type for a specific operation.
//===----------------------------------------------------------------------===//
template <typename Op>
struct ScalarOp {
using FOp = void;
using IOp = void;
};
template <typename FOp>
using ScalarFOp = typename ScalarOp<FOp>::FOp;
template <typename IOp>
using ScalarIOp = typename ScalarOp<IOp>::IOp;
// Get the identity element of a operation.
// Return NULL if the function does not have identity.
template <typename DataType, typename Op>
DataType getIdentityValue() {
return NULL;
}
//===----------------------------------------------------------------------===//
// This is used in the innermost loop of a KrnlIterateOp to insert computation
// composed of one or many scalar ops.
// Use template specialization for each of different ONNX operations.
//===----------------------------------------------------------------------===//
template <typename Op>
Value mapToLowerScalarOp(Operation *op, ArrayRef<Type> result_types,
ArrayRef<Value> operands,
ConversionPatternRewriter &rewriter) {
auto loc = op->getLoc();
Type element_type = operands.front().getType();
if (element_type.isa<IntegerType>()) {
return rewriter.create<ScalarIOp<Op>>(loc, result_types, operands,
mlir::None);
} else if (element_type.isa<FloatType>()) {
return rewriter.create<ScalarFOp<Op>>(loc, result_types, operands,
mlir::None);
} else {
emitError(loc, "unsupported element type");
return nullptr;
}
}
//===----------------------------------------------------------------------===//
// Conversion from Tensor type to the Standard dialect MemRef type.
//===----------------------------------------------------------------------===//
struct TensorTypeConverter : public TypeConverter {
using TypeConverter::TypeConverter;
TensorTypeConverter() {
addConversion(convertType);
}
static LogicalResult convertType(Type t, SmallVectorImpl<Type> &results) {
if (auto type = convertToMemRefType(t)) {
results.push_back(type);
return success();
}
results.push_back(t);
return success();
}
/// Return true if the inputs and outputs of the given function type are
/// legal. [Taken from MLIR and adapted to only check the legality of the
/// inputs. Once unranked results can be handled gracefully this
/// override needs to be removed in favour of the original MLIR one.]
bool isSignatureLegal(FunctionType funcType) {
return llvm::all_of(funcType.getInputs(),
[this](Type type) { return isLegal(type); });
}
};
//===----------------------------------------------------------------------===//
// Functions to add lowering patterns for frontend operations.
//===----------------------------------------------------------------------===//
// `math` directory methods:
void populateLoweringONNXElementwiseOpPattern(
OwningRewritePatternList &patterns, MLIRContext *ctx);
void populateLoweringONNXGemmOpPattern(OwningRewritePatternList &patterns,
MLIRContext *ctx);
void populateLoweringONNXMatMulOpPattern(
OwningRewritePatternList &patterns, MLIRContext *ctx);
void populateLoweringONNXReductionOpPattern(
OwningRewritePatternList &patterns, MLIRContext *ctx);
void populateLoweringONNXSoftmaxOpPattern(
OwningRewritePatternList &patterns, MLIRContext *ctx);
// `nn` directory methods:
void populateLoweringONNXConvOpPattern(
OwningRewritePatternList &patterns, MLIRContext *ctx);
void populateLoweringONNXNormalizationOpPattern(
OwningRewritePatternList &patterns, MLIRContext *ctx);
// `tensor` directory methods:
void populateLoweringONNXUnsqueezeOpPattern(
OwningRewritePatternList &patterns, MLIRContext *ctx);
void populateLoweringONNXTransposeOpPattern(
OwningRewritePatternList &patterns, MLIRContext *ctx);
void populateLoweringONNXReshapeOpPattern(
OwningRewritePatternList &patterns, MLIRContext *ctx);
void populateLoweringONNXIdentityOpPattern(
OwningRewritePatternList &patterns, MLIRContext *ctx);

View File

@ -1,4 +1,4 @@
//===----- identity.inc - Lowering Identity Op ----------------------------===//
//===----- identity.cpp - Lowering Identity Op ----------------------------===//
//
// Copyright 2019 The IBM Research Authors.
//
@ -8,6 +8,10 @@
//
//===----------------------------------------------------------------------===//
#include "src/conversion/onnx_to_krnl/onnx_to_krnl_common.hpp"
using namespace mlir;
struct ONNXIdentityOpLowering : public ConversionPattern {
ONNXIdentityOpLowering(MLIRContext *ctx)
: ConversionPattern(mlir::ONNXIdentityOp::getOperationName(), 1, ctx) {}

View File

@ -1,4 +1,4 @@
//===----- reshape.inc - Lowering Reshape Op ------------------------------===//
//===----- reshape.cpp - Lowering Reshape Op ------------------------------===//
//
// Copyright 2019 The IBM Research Authors.
//
@ -8,6 +8,10 @@
//
//===----------------------------------------------------------------------===//
#include "src/conversion/onnx_to_krnl/onnx_to_krnl_common.hpp"
using namespace mlir;
struct ONNXReshapeOpLowering : public ConversionPattern {
ONNXReshapeOpLowering(MLIRContext *ctx)
: ConversionPattern(mlir::ONNXReshapeOp::getOperationName(), 1, ctx) {}

View File

@ -1,4 +1,4 @@
//===----- transpose.inc - Lowering Transpose Op --------------------------===//
//===----- transpose.cpp - Lowering Transpose Op --------------------------===//
//
// Copyright 2019 The IBM Research Authors.
//
@ -8,6 +8,10 @@
//
//===----------------------------------------------------------------------===//
#include "src/conversion/onnx_to_krnl/onnx_to_krnl_common.hpp"
using namespace mlir;
struct ONNXTransposeOpLowering : public ConversionPattern {
ONNXTransposeOpLowering(MLIRContext *ctx)
: ConversionPattern(mlir::ONNXTransposeOp::getOperationName(), 1, ctx) {}

View File

@ -1,4 +1,4 @@
//===----- unsqueeze.inc - Lowering Unsqueeze Op --------------------------===//
//===----- unsqueeze.cpp - Lowering Unsqueeze Op --------------------------===//
//
// Copyright 2019 The IBM Research Authors.
//
@ -8,6 +8,10 @@
//
//===----------------------------------------------------------------------===//
#include "src/conversion/onnx_to_krnl/onnx_to_krnl_common.hpp"
using namespace mlir;
struct ONNXUnsqueezeOpLowering : public ConversionPattern {
ONNXUnsqueezeOpLowering(MLIRContext *ctx)
: ConversionPattern(mlir::ONNXUnsqueezeOp::getOperationName(), 1, ctx) {}