torch-mlir/lib/Conversion/TorchToLinalg/IndirectDataMovement.cpp

1103 lines
46 KiB
C++

//===----------------------------------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
// Also available under a BSD-style license. See LICENSE.
//
//===----------------------------------------------------------------------===//
#include "torch-mlir/Conversion/TorchToLinalg/TorchToLinalg.h"
#include "../PassDetail.h"
#include "PopulatePatterns.h"
#include "Utils.h"
#include "mlir/Dialect/Arith/IR/Arith.h"
#include "mlir/Dialect/ControlFlow/IR/ControlFlowOps.h"
#include "mlir/Dialect/Linalg/IR/Linalg.h"
#include "mlir/Dialect/Math/IR/Math.h"
#include "mlir/Dialect/Tensor/IR/Tensor.h"
#include "mlir/IR/Matchers.h"
#include "torch-mlir/Conversion/Utils/Utils.h"
#include "torch-mlir/Dialect/Torch/IR/TorchDialect.h"
#include "torch-mlir/Dialect/Torch/IR/TorchOps.h"
#include "torch-mlir/Dialect/Torch/Utils/TorchUpstream.h"
#include "torch-mlir/Dialect/Torch/Utils/Utils.h"
using namespace mlir;
using namespace mlir::torch;
using namespace mlir::torch::Torch;
static void createLinalgPayloadCalculationForGatherOps(
OpBuilder &b, Location loc, Value input, int64_t inputRank, Value index,
int64_t dim, int64_t outputRank) {
SmallVector<Value> indices;
for (int i = 0; i < inputRank; i++) {
if (i == dim) {
indices.push_back(castIntToIndex(b, loc, index));
} else {
// `outputRank` might be larger than `inputRank`. The `linalg::IndexOp`
// takes in the dimension of the output. Add `inputDimOffset` to
// related to the correct dimension of the output for dimension larger
// than the given `dim`.
int64_t inputDimOffset = i < dim ? 0 : outputRank - inputRank;
indices.push_back(b.create<linalg::IndexOp>(loc, i + inputDimOffset));
}
}
// Assert index < input.sizes[dim]
Value indexLTInputDim = b.create<arith::CmpIOp>(
loc, arith::CmpIPredicate::slt, castIntToIndex(b, loc, index),
getDimOp(b, loc, input, dim));
b.create<cf::AssertOp>(
loc, indexLTInputDim,
b.getStringAttr("index must be smaller than dim size"));
// Assert index >= 0
Value cst0 = b.create<arith::ConstantOp>(loc, b.getZeroAttr(index.getType()));
Value indexGEThanZero =
b.create<arith::CmpIOp>(loc, arith::CmpIPredicate::sge, index, cst0);
b.create<cf::AssertOp>(loc, indexGEThanZero,
b.getStringAttr("index must be larger or equal to 0"));
Value extract = b.create<tensor::ExtractOp>(loc, input, indices);
b.create<linalg::YieldOp>(loc, extract);
}
namespace {
class ConvertAtenGatherOp : public OpConversionPattern<AtenGatherOp> {
public:
using OpConversionPattern::OpConversionPattern;
LogicalResult
matchAndRewrite(AtenGatherOp op, OpAdaptor adaptor,
ConversionPatternRewriter &rewriter) const override {
if (failed(verifyLinalgCompatibleTypes(op, rewriter)))
return failure();
Location loc = op->getLoc();
Value dimValue = op.dim();
int64_t dim;
if (!matchPattern(dimValue, m_TorchConstantInt(&dim)))
return op.emitError("unimplemented: dim is not constant");
Value indices = adaptor.index();
Value self = adaptor.self();
RankedTensorType newResultTy =
getTypeConverter()->convertType(op.getType()).cast<RankedTensorType>();
int64_t rank = newResultTy.getRank();
SmallVector<Value> sizes = getTensorSizes(rewriter, loc, indices);
Value result = createZeroInitTensor(rewriter, loc, sizes,
newResultTy.getElementType());
SmallVector<AffineMap, 2> affineMaps(2,
rewriter.getMultiDimIdentityMap(rank));
SmallVector<StringRef> iteratorTypes(rank, getParallelIteratorTypeName());
auto genericOp = rewriter
.create<linalg::GenericOp>(
loc, result.getType(), indices, result, affineMaps,
iteratorTypes,
[&](OpBuilder &b, Location loc, ValueRange args) {
auto index = args[0];
createLinalgPayloadCalculationForGatherOps(
b, loc, self, rank, index, dim, rank);
})
.getResult(0);
rewriter.replaceOpWithNewOp<tensor::CastOp>(op, newResultTy, genericOp);
return success();
}
};
} // namespace
namespace {
class ConvertAtenEmbeddingOp : public OpConversionPattern<AtenEmbeddingOp> {
public:
using OpConversionPattern::OpConversionPattern;
LogicalResult
matchAndRewrite(AtenEmbeddingOp op, OpAdaptor adaptor,
ConversionPatternRewriter &rewriter) const override {
if (failed(verifyLinalgCompatibleTypes(op, rewriter)))
return failure();
Location loc = op->getLoc();
Value weight = adaptor.weight();
Value indices = adaptor.indices();
RankedTensorType newResultType =
typeConverter->convertType(op.getType()).cast<RankedTensorType>();
auto weightTy = weight.getType().cast<RankedTensorType>();
if (weightTy.getRank() != 2)
return rewriter.notifyMatchFailure(op, "weight must be rank 2");
Value embeddingDim = getDimOp(rewriter, loc, weight, 1);
Type elemTy = weightTy.getElementType();
SmallVector<Value> sizes = getTensorSizes(rewriter, loc, indices);
sizes.push_back(embeddingDim);
int64_t resultRank = sizes.size();
auto indicesTy = indices.getType().cast<RankedTensorType>();
int64_t indicesRank = indicesTy.getRank();
SmallVector<AffineExpr> indicesExprs;
for (int i = 0; i < indicesRank; i++)
indicesExprs.push_back(rewriter.getAffineDimExpr(i));
auto indicesAffineMap = AffineMap::get(
/*dimCount=*/resultRank,
/*symbolCount=*/0, indicesExprs, op->getContext());
SmallVector<AffineMap, 2> indexingMaps = {
indicesAffineMap,
rewriter.getMultiDimIdentityMap(resultRank),
};
SmallVector<StringRef> iteratorTypes(sizes.size(),
getParallelIteratorTypeName());
Value initTensor =
rewriter.create<tensor::EmptyOp>(loc, getAsOpFoldResult(sizes), elemTy);
Value embeddingResult =
rewriter
.create<linalg::GenericOp>(
loc, initTensor.getType(), indices, initTensor,
/*indexingMaps=*/indexingMaps, /*iteratorTypes=*/iteratorTypes,
[&](OpBuilder &b, Location loc, ValueRange args) {
Value index = args[0];
createLinalgPayloadCalculationForGatherOps(
b, loc, weight, weightTy.getRank(), index, /*dim=*/0,
resultRank);
})
.getResult(0);
rewriter.replaceOpWithNewOp<tensor::CastOp>(op, newResultType,
embeddingResult);
return success();
}
};
} // namespace
namespace {
// AtenEmbeddingPaddingIdxOp
// SUM mode == integer 0
// Sums bags of embeddings together from a weight tensor based on an index and
// offset Vector. Example arguments weight = [[1, 3, 5, 3],
// [3, 4, 2, 1],
// [2, 2, 3, 2],
// [0, 4, 2, 1]]
//
// indices = [0, 2, 3, 1, 2, 3, 2, 1, 0, 1]
// offsets = [0, 3, 5]
//
// output_tensor = initZeroTensor(offsets_length, embedding_size)
//
// for i in range(offsets_length): <- dim0
// for j in range(indices_length): <- dim1
// for k in range(embedding_size): <- dim2
// if(offsets[i] <= j and j < offsets[i+1]):
// output_tensor[i][k] = output_tensor[i][k] +
// weight[indices[j]][k]
// else:
// break
//
// Indexing maps for linalg::Generic ops
//
//
// indices_indexing_map = (d0, d1, d2) -> (d1)
// offset_indexing_map = (d0, d1, d2) -> (d0)
// output_indexing_map = (d0, d1, d2) -> (d0, d2)
//
// TODO: Find an optimal lowering.
// current lowering is not optimal for bags of large embeddings.
// Since it traverses the output tensor multiple times.
//
//
class ConvertAtenEmbeddingBagPaddingIdxOp
: public OpConversionPattern<AtenEmbeddingBagPaddingIdxOp> {
public:
using OpConversionPattern::OpConversionPattern;
LogicalResult
matchAndRewrite(AtenEmbeddingBagPaddingIdxOp op, OpAdaptor adaptor,
ConversionPatternRewriter &rewriter) const override {
if (failed(verifyLinalgCompatibleTypes(op, rewriter)))
return failure();
Location loc = op->getLoc();
auto context = op->getContext();
Value weight = adaptor.weight();
Value indices = adaptor.indices();
Value offsets = adaptor.offsets();
Value scaleGradByFreq = op.scale_grad_by_freq();
Value mode = op.mode();
Value sparse = op.sparse();
Value includeLastOffset = op.include_last_offset();
bool scaleGradByFreqBool;
if (!matchPattern(scaleGradByFreq,
m_TorchConstantBool(&scaleGradByFreqBool))) {
return rewriter.notifyMatchFailure(
op, "scale_grad_by_freq is expected to be a constant boolean value.");
}
if (scaleGradByFreqBool) {
return rewriter.notifyMatchFailure(
op, "Unimplemented: scale_grad_by_freq=True.");
}
int64_t modeInt;
if (!matchPattern(mode, m_TorchConstantInt(&modeInt))) {
return rewriter.notifyMatchFailure(
op, "mode is expected to be a constant integer value.");
}
if (modeInt != torch_upstream::EmbeddingBagMode::MODE_SUM) {
return rewriter.notifyMatchFailure(
op,
"Unimplemented: Mean and Max mode are not supported yet for EmbeddingBag.");
}
bool isSparse;
if (!matchPattern(sparse, m_TorchConstantBool(&isSparse))) {
return rewriter.notifyMatchFailure(
op, "sparse is expected to be a constant boolean value.");
}
if (isSparse) {
return rewriter.notifyMatchFailure(
op,
"Unimplemented: Sparse mode is not supported yet for EmbeddingBag.");
}
bool discardLastOffset;
if (!matchPattern(includeLastOffset,
m_TorchConstantBool(&discardLastOffset))) {
return rewriter.notifyMatchFailure(
op,
"include_last_offset is expected to be a constant boolean value.");
}
auto weightTy = weight.getType().cast<RankedTensorType>();
if (weightTy.getRank() != 2)
return rewriter.notifyMatchFailure(op, "weight must be rank 2");
auto indicesTy = indices.getType().cast<RankedTensorType>();
if (indicesTy.getRank() != 1)
return rewriter.notifyMatchFailure(op, "indices must be a vector");
auto offsetsTy = offsets.getType().cast<RankedTensorType>();
if (offsetsTy.getRank() != 1)
return rewriter.notifyMatchFailure(op, "offsets much be a vector");
Type weightElemTy = weightTy.getElementType();
int64_t iterationMapDimension = weightTy.getRank() + indicesTy.getRank();
SmallVector<AffineExpr> indicesExpr;
indicesExpr.push_back(mlir::getAffineDimExpr(1, context));
auto indicesIndexingMap =
AffineMap::get(/*dimCount=*/iterationMapDimension, /*symbolCount=*/0,
indicesExpr, context);
SmallVector<AffineExpr> offsetsExpr;
offsetsExpr.push_back(mlir::getAffineDimExpr(0, context));
auto offsetIndexingMap =
AffineMap::get(/*dimCount=*/iterationMapDimension, /*symbolCount=*/0,
offsetsExpr, context);
SmallVector<AffineExpr> outputExpr;
outputExpr.push_back(mlir::getAffineDimExpr(0, context));
outputExpr.push_back(mlir::getAffineDimExpr(2, context));
auto outputIndexingMap =
AffineMap::get(/*dimCount=*/iterationMapDimension, /*symbolCount=*/0,
outputExpr, context);
SmallVector<AffineMap, 3> indexingMaps = {
indicesIndexingMap,
offsetIndexingMap,
outputIndexingMap,
};
// Reduce along the indices dim
SmallVector<StringRef> iteratorTypes({getParallelIteratorTypeName(),
getReductionIteratorTypeName(),
getParallelIteratorTypeName()});
Value embeddingDim = getDimOp(rewriter, loc, weight, 1);
Value initTensor;
Value offsetsLength;
Value indicesLength;
if (!discardLastOffset) {
SmallVector<Value> sizes{getDimOp(rewriter, loc, offsets, 0),
embeddingDim};
initTensor = createZeroInitTensor(rewriter, loc, sizes, weightElemTy);
offsetsLength = getDimOp(rewriter, loc, offsets, 0);
indicesLength = getDimOp(rewriter, loc, indices, 0);
} else {
return rewriter.notifyMatchFailure(
op, "Unimplemented: include last offset is not yet "
"supported for EmbeddingBag.");
}
Value embeddingBagResult =
rewriter
.create<linalg::GenericOp>(
loc, initTensor.getType(), ValueRange{indices, offsets},
initTensor,
/*indexingMaps=*/indexingMaps,
/*iteratorTypes=*/iteratorTypes,
[&](OpBuilder &b, Location loc, ValueRange args) {
Value indexInIndices = args[0];
Value offsetsI = args[1];
Value initTensorElem = args[2];
Value indexI = b.create<linalg::IndexOp>(loc, /*value=*/0);
Value indexIToInt = castIndexToInt64(b, loc, indexI);
Value one = getConstant(
b, loc, 1,
mlir::IntegerType::get(getContext(), 64,
IntegerType::Signless));
Value offsetIndexPlusOneInt =
b.create<arith::AddIOp>(loc, indexIToInt, one);
Value offsetIndexPlusOne =
castIntToIndex(b, loc, offsetIndexPlusOneInt);
Value checkLast = b.create<arith::CmpIOp>(
loc, arith::CmpIPredicate::eq,
castIndexToInt64(b, loc, offsetsLength),
offsetIndexPlusOneInt);
Value nextOffset = b.create<arith::SelectOp>(
loc, checkLast, castIndexToInt64(b, loc, indicesLength),
b.create<tensor::ExtractOp>(loc, offsets,
offsetIndexPlusOne));
Value indicesIndex = castIndexToInt64(
b, loc, b.create<linalg::IndexOp>(loc, /*value=*/1));
Value offsetLessThanIndicesIndex = b.create<arith::CmpIOp>(
loc, arith::CmpIPredicate::slt, offsetsI, indicesIndex);
Value offsetEqualToIndicesIndex = b.create<arith::CmpIOp>(
loc, arith::CmpIPredicate::eq, offsetsI, indicesIndex);
Value offsetLessThanOrEqualToIndicesIndex =
b.create<arith::OrIOp>(loc, offsetLessThanIndicesIndex,
offsetEqualToIndicesIndex);
Value indicesIndexLessThanNextOffset =
b.create<arith::CmpIOp>(loc, arith::CmpIPredicate::slt,
indicesIndex, nextOffset);
Value indicesIndexWithinBounds = b.create<arith::AndIOp>(
loc, offsetLessThanOrEqualToIndicesIndex,
indicesIndexLessThanNextOffset);
SmallVector<Value> indexIntoWeight;
indexIntoWeight.push_back(
castIntToIndex(b, loc, indexInIndices));
indexIntoWeight.push_back(
b.create<linalg::IndexOp>(loc, /*value=*/2));
Value weightElem = b.create<tensor::ExtractOp>(
loc, weight, indexIntoWeight);
Value addResult = b.create<arith::AddFOp>(loc, weightElem,
initTensorElem);
Value select =
b.create<arith::SelectOp>(loc, indicesIndexWithinBounds,
addResult, initTensorElem);
b.create<linalg::YieldOp>(loc, select);
})
.getResult(0);
// cast outputType.
auto restulType0 = typeConverter->convertType(op->getResult(0).getType());
Value castedEmbeddingBagResult =
rewriter.create<tensor::CastOp>(loc, restulType0, embeddingBagResult);
// offset2 tensor, this should be an empty tensor for the sum mode
SmallVector<Value> offsetResultSize;
Type offsetElemTy = offsetsTy.getElementType();
Value zeroDim = rewriter.create<arith::ConstantIndexOp>(loc, /*value=*/0);
offsetResultSize.push_back(zeroDim);
Value offsetResult = rewriter.create<tensor::EmptyOp>(
loc, getAsOpFoldResult(offsetResultSize), offsetElemTy);
auto resultType1 = typeConverter->convertType(op->getResult(1).getType());
Value castedOffsetResult =
rewriter.create<tensor::CastOp>(loc, resultType1, offsetResult);
SmallVector<Value> offsetSize = getTensorSizes(rewriter, loc, offsets);
// bagsize, vector of size offset with zeros, I think this is always just
// a vector of zeros in the sum mode
Value bagSize =
createZeroInitTensor(rewriter, loc, offsetSize, offsetElemTy);
auto resultType2 = typeConverter->convertType(op->getResult(2).getType());
Value castedBagSizeResult =
rewriter.create<tensor::CastOp>(loc, resultType2, bagSize);
// max indices, vector of size offset with zeros, this is also always a
// vector of zeros in the sum mode. Its mainly used in the max mode.
Value indicesOut =
createZeroInitTensor(rewriter, loc, offsetSize, offsetElemTy);
auto resultType3 = typeConverter->convertType(op->getResult(3).getType());
Value castedMaxIndices =
rewriter.create<tensor::CastOp>(loc, resultType3, indicesOut);
rewriter.replaceOp(op, {castedEmbeddingBagResult, castedOffsetResult,
castedBagSizeResult, castedMaxIndices});
return success();
}
};
} // namespace
namespace {
// Let's say we have an input tensor: initialized with some random values of
// size [4, 5, 6]. An index tensor (always 1-d): [0, 2] of size [2], and an
// integer argument dim = 1. The size of the output tensor will be [4, 2, 6].
// The approach is as follows:
//
// for i in range(input.size[0])
// for j in range(index.size[0])
// for k in range(input.size[2])
// indexValue = index[j]
// output[i,j,k] = input[i,indexValue,k]
class ConvertAtenIndexSelectOp : public OpConversionPattern<AtenIndexSelectOp> {
public:
using OpConversionPattern::OpConversionPattern;
LogicalResult
matchAndRewrite(AtenIndexSelectOp op, OpAdaptor adaptor,
ConversionPatternRewriter &rewriter) const override {
if (failed(verifyLinalgCompatibleTypes(op, rewriter)))
return failure();
Location loc = op.getLoc();
Value input = adaptor.self();
Value indices = adaptor.index();
RankedTensorType inputType = input.getType().cast<RankedTensorType>();
RankedTensorType resultType = getTypeConverter()
->convertType(op->getResult(0).getType())
.cast<RankedTensorType>();
Type elementType = resultType.getElementType();
unsigned inputRank = inputType.getRank();
int64_t dimInt;
if (!matchPattern(op.dim(), m_TorchConstantInt(&dimInt)))
return op->emitError("unimplemented: dim is not constant");
SmallVector<Value> resultShape = getTensorSizes(rewriter, loc, input);
resultShape[dimInt] = getTensorSizes(rewriter, loc, indices)[0];
Value initTensor = rewriter.create<tensor::EmptyOp>(
loc, getAsOpFoldResult(resultShape), elementType);
SmallVector<AffineExpr> resultExpr;
AffineExpr indicesExpr = rewriter.getAffineDimExpr(dimInt);
SmallVector<StringRef> iteratorTypes;
for (unsigned i = 0; i < inputRank; i++) {
resultExpr.push_back(rewriter.getAffineDimExpr(i));
iteratorTypes.push_back(getParallelIteratorTypeName());
}
auto indexingMaps = AffineMap::inferFromExprList({indicesExpr, resultExpr});
Value finalRes =
rewriter
.create<linalg::GenericOp>(
loc, initTensor.getType(), ValueRange{indices}, initTensor,
/*indexingMaps=*/indexingMaps,
/*iteratorTypes=*/iteratorTypes,
[&](OpBuilder &b, Location loc, ValueRange args) {
Value index = rewriter.create<arith::IndexCastOp>(
loc, rewriter.getIndexType(), args[0]);
SmallVector<Value> indexTarget;
for (unsigned i = 0; i < inputRank; i++)
indexTarget.push_back(b.create<linalg::IndexOp>(loc, i));
indexTarget[dimInt] = index;
Value extractedElement =
b.create<tensor::ExtractOp>(loc, input, indexTarget);
b.create<linalg::YieldOp>(loc, extractedElement);
})
.getResult(0);
rewriter.replaceOpWithNewOp<tensor::CastOp>(op, resultType, finalRes);
return success();
}
};
} // namespace
// IndexTensor for multiple input tensors broadcasts their shapes to a common
// shape and then replaces the indexed dims with the indices given by the
// indexing tensors:
// x[i_1, i_2, ..., i_M] = result
// result[...] = x[i_1[...], i_2[...], ..., i_M[...]]
//
// where the result shape is computed as follows:
// 1. broadcast i_1, i_2, ..., i_M to a common shape
// 2. if i_1, i_2, ..., i_M is not contiguous, transpose the broadcasted
// shape to the beginning of the result shape, while removing the
// unchanged dims (marked by None)
// 3. Otherwise replace the indexed dims with the broadcasted shape
//
// e.g. x: [2, 3]
// x[[4], [6, 1]] -> x[6, 4]
namespace {
class ConvertAtenIndexTensorOp : public OpConversionPattern<AtenIndexTensorOp> {
public:
using OpConversionPattern::OpConversionPattern;
LogicalResult
matchAndRewrite(AtenIndexTensorOp op, OpAdaptor adaptor,
ConversionPatternRewriter &rewriter) const override {
if (failed(verifyLinalgCompatibleTypes(op, rewriter)))
return failure();
Location loc = op.getLoc();
Value input = adaptor.self();
Value indices = op.indices();
SmallVector<Value> indicesTuple;
if (!getListConstructElements(indices, indicesTuple)) {
return rewriter.notifyMatchFailure(
op, "unimplemented: the indices list is not from a list construct");
}
SmallVector<Value> indicesVal =
getTypeConvertedValues(rewriter, loc, getTypeConverter(), indicesTuple);
// Identify the indices with non-None index tensors and determine if they
// are contiguous within the input list.
SmallVector<int> indexTensorDims;
SmallVector<Value> indexTensors;
bool contiguous = true;
for (auto i : llvm::seq(0, (int)indicesVal.size())) {
Value index = indicesVal[i];
if (!index || failed(checkNotNone(rewriter, op, index)))
continue;
if (!indexTensorDims.empty() && indexTensorDims.back() != i - 1)
contiguous = false;
indexTensorDims.push_back(i);
indexTensors.push_back(index);
}
if (indexTensors.empty()) {
return rewriter.notifyMatchFailure(
op, "aten.index.Tensor: index tensor must not be None");
}
RankedTensorType inputType = input.getType().cast<RankedTensorType>();
RankedTensorType resultType = getTypeConverter()
->convertType(op->getResult(0).getType())
.cast<RankedTensorType>();
Type elementType = resultType.getElementType();
int inputRank = inputType.getRank();
int resultRank = resultType.getRank();
int firstIndexDim = indexTensorDims[0];
int replacedIndexCount = indexTensorDims.size();
int64_t startIndex = contiguous ? firstIndexDim : 0;
// Currently we only support statically sized index tensors or dynamic size
// index tensors without overlapping dynamic dims when there is more than
// one index tensor.
// TODO: Add support for dynamic size index tensors with overlapping
// dynamic dims.
SmallVector<Value> broadcastedIndexShape;
if (indexTensors.size() > 1) {
int maxRank = -1;
for (auto indexTensor : indexTensors) {
RankedTensorType indexTensorType =
indexTensor.getType().cast<RankedTensorType>();
maxRank = std::max(maxRank, (int)indexTensorType.getRank());
}
// Because we are assuming static shapes, we can get the shape of the
// broadcasted index tensors from the shape refinement pass
auto refinedResultShape = resultType.getShape();
for (auto i : llvm::seq(startIndex, startIndex + maxRank)) {
auto resultDimSize = refinedResultShape[i];
if (ShapedType::isDynamic(resultDimSize)) {
SmallVector<Value> dynamicDims;
int64_t staticDimSize = -1;
for (auto indexTensor : indexTensors) {
RankedTensorType indexTensorType =
indexTensor.getType().cast<RankedTensorType>();
int64_t indexTensorRank = indexTensorType.getRank();
if ((maxRank - indexTensorRank) > (i - startIndex))
continue;
int64_t dim = i - startIndex - maxRank + indexTensorRank;
if (ShapedType::isDynamic(indexTensorType.getShape()[dim]))
dynamicDims.push_back(getDimOp(rewriter, loc, indexTensor, dim));
else
staticDimSize =
std::max(staticDimSize, indexTensorType.getShape()[dim]);
}
if (dynamicDims.size() >= 2)
return rewriter.notifyMatchFailure(
op,
"unimplemented: index tensors with overlapping dynamic dims");
if (staticDimSize > 1) {
Value cstStaticDimSize = getConstant(rewriter, loc, staticDimSize,
rewriter.getIndexType());
auto equalToRunning = rewriter.create<arith::CmpIOp>(
loc, arith::CmpIPredicate::eq, cstStaticDimSize,
dynamicDims[0]);
rewriter.create<cf::AssertOp>(loc, equalToRunning,
"mismatched size for broadcast");
}
broadcastedIndexShape.push_back(dynamicDims[0]);
} else {
broadcastedIndexShape.push_back(getConstant(
rewriter, loc, resultDimSize, rewriter.getIndexType()));
}
}
} else {
// For a single indexing tensor we can simply use its (dynamic) sizes
broadcastedIndexShape =
getTensorSizes(rewriter, loc, indexTensors.front());
}
// This result shape calculation assumes that there is only one
// index tensor, or all of the index tensors are statically shaped.
int broadcastRank = broadcastedIndexShape.size();
SmallVector<Value> resultShape;
if (contiguous) {
for (auto i : llvm::seq(0, firstIndexDim)) {
resultShape.push_back(getDimOp(rewriter, loc, input, i));
}
resultShape.append(broadcastedIndexShape);
for (auto i : llvm::seq((int)resultShape.size(), resultRank)) {
resultShape.push_back(getDimOp(rewriter, loc, input,
i - broadcastRank + replacedIndexCount));
}
} else {
resultShape.append(broadcastedIndexShape);
int j = 0;
for (auto i : llvm::seq(0, inputRank)) {
if (j < replacedIndexCount && i == indexTensorDims[j]) {
j++;
continue;
}
resultShape.push_back(getDimOp(rewriter, loc, input, i));
}
}
// Initialize the indexing maps for the generic op. Because we are assuming
// static shapes for the indexing tensors when there are more than 1, we can
// safely map all size 1 dims to 0 in the corresponding affine maps.
// TODO: For dynamic shapes, we have to either broadcast the index tensors
// to a common shape or introduce some form of control flow.
Value initTensor = rewriter.create<tensor::EmptyOp>(
loc, getAsOpFoldResult(resultShape), elementType);
SmallVector<AffineMap> indexingMaps;
SmallVector<StringRef> iteratorTypes;
for (auto indexTensor : indexTensors) {
RankedTensorType indexTensorType =
indexTensor.getType().cast<RankedTensorType>();
auto indexTensorShape = indexTensorType.getShape();
int rank = indexTensorShape.size();
SmallVector<AffineExpr> indicesExpr;
for (auto dim : llvm::seq(0, rank)) {
if (indexTensorShape[dim] == 1) {
indicesExpr.push_back(rewriter.getAffineConstantExpr(0));
continue;
}
indicesExpr.push_back(
rewriter.getAffineDimExpr(startIndex + broadcastRank - rank + dim));
}
indexingMaps.push_back(
AffineMap::get(resultRank, 0, indicesExpr, op->getContext()));
}
SmallVector<AffineExpr> resultExpr;
for (auto i : llvm::seq(0, resultRank)) {
resultExpr.push_back(rewriter.getAffineDimExpr(i));
iteratorTypes.push_back(getParallelIteratorTypeName());
}
indexingMaps.push_back(
AffineMap::get(resultRank, 0, resultExpr, op->getContext()));
Value finalRes =
rewriter
.create<linalg::GenericOp>(
loc, initTensor.getType(), indexTensors, initTensor,
indexingMaps, iteratorTypes,
[&](OpBuilder &b, Location loc, ValueRange args) {
SmallVector<Value> extractionIndices;
if (contiguous) {
for (auto i : llvm::seq(0, firstIndexDim)) {
extractionIndices.push_back(
b.create<linalg::IndexOp>(loc, i));
}
for (auto i : llvm::seq(0, (int)indexTensorDims.size())) {
extractionIndices.push_back(
castIntToIndex(b, loc, args[i]));
}
for (auto i :
llvm::seq((int)extractionIndices.size(), inputRank)) {
extractionIndices.push_back(b.create<linalg::IndexOp>(
loc, i + broadcastRank - replacedIndexCount));
}
} else {
int indexCount = 0, unchanged = 0;
for (auto i : llvm::seq(0, inputRank)) {
if (indexCount < replacedIndexCount &&
i == indexTensorDims[indexCount]) {
extractionIndices.push_back(
castIntToIndex(b, loc, args[indexCount++]));
continue;
}
extractionIndices.push_back(b.create<linalg::IndexOp>(
loc, broadcastRank + unchanged));
unchanged++;
}
}
Value extractedElement = b.create<tensor::ExtractOp>(
loc, input, extractionIndices);
b.create<linalg::YieldOp>(loc, extractedElement);
})
.getResult(0);
rewriter.replaceOpWithNewOp<tensor::CastOp>(op, resultType, finalRes);
return success();
}
};
} // namespace
// `getScaledDims` scales the `dim` value with a scale factor `ScaleFactor`.
// The `dim` and `scaleFactor` are assumed to be of index and float type
// respectively. `scaledDim = int(floor(float(dim) * scaleFactor))`.
static Value getScaledDims(OpBuilder &builder, Location loc, Value dim,
Value scaleFactor) {
Value dimInt = castIndexToInt64(builder, loc, dim);
Value dimFp =
builder.create<arith::SIToFPOp>(loc, scaleFactor.getType(), dimInt);
Value scaleDim = builder.create<arith::MulFOp>(loc, dimFp, scaleFactor);
Value floorDim = builder.create<math::FloorOp>(loc, scaleDim);
Value scaledDimToIndex = castIntToIndex(
builder, loc,
builder.create<arith::FPToSIOp>(loc, dimInt.getType(), floorDim));
return scaledDimToIndex;
}
// `getScaleFactor` returns the scale factor from input to output dimension.
// The `dim` and `scaledDim` are assumed to be of index and int64 type
// respectively. scale_factor = (scaled_dim // dim).
static Value getScaleFactor(OpBuilder &builder, Location loc, Value dim,
Value scaledDim) {
Value dimInt = castIndexToInt64(builder, loc, dim);
Value scaleFactorInt =
builder.create<arith::CeilDivSIOp>(loc, scaledDim, dimInt);
return scaleFactorInt;
}
// N, C, H, W = input_tensor.shape
// N, C, H_scaled, W_scaled = out_tensor.shape
// H_factor, W_factor = H_scaled/H, W_scaled/W
// for i in range(N):
// for j in range(C):
// for k in range(H_scaled):
// for l in range(W_scaled):
// out_tensor[i, j, k, l] = input[i, j, k//H_factor, l//W_factor]
namespace {
class ConvertAtenUpsampleNearest2dVecOp
: public OpConversionPattern<AtenUpsampleNearest2dVecOp> {
public:
using OpConversionPattern::OpConversionPattern;
LogicalResult
matchAndRewrite(AtenUpsampleNearest2dVecOp op, OpAdaptor adaptor,
ConversionPatternRewriter &rewriter) const override {
Location loc = op->getLoc();
Value input = adaptor.input();
Type resultType = getTypeConverter()->convertType(op.getResult().getType());
auto inputType = input.getType().cast<RankedTensorType>();
auto inputRank = inputType.getRank();
Type elementType = inputType.getElementType();
SmallVector<Value> dims = getTensorSizes(rewriter, loc, input);
SmallVector<Value, 2> scaleFactorsInt;
// The dimension at which the scaling starts.
unsigned hDimOffset = 2;
if (!adaptor.scale_factors().getType().isa<Torch::NoneType>()) {
SmallVector<Value, 2> scaleFactorsTorchFloat;
if (!getListConstructElements(op.scale_factors(), scaleFactorsTorchFloat))
return rewriter.notifyMatchFailure(
op, "unimplemented: the scale_factors is not constructed from "
"ListConstruct");
SmallVector<Value, 2> scaleFactorsFloatValues;
scaleFactorsFloatValues = getTypeConvertedValues(
rewriter, loc, getTypeConverter(), scaleFactorsTorchFloat);
// Convert float values to int values.
// int_value = (int64_t)ceil(float_value)
for (auto floatValue : scaleFactorsFloatValues) {
Value ceilVal = rewriter.create<math::CeilOp>(loc, floatValue);
Value intVal = rewriter.create<arith::FPToSIOp>(
loc, rewriter.getI64Type(), ceilVal);
scaleFactorsInt.push_back(intVal);
}
for (unsigned i = 0; i < scaleFactorsFloatValues.size(); i++)
dims[hDimOffset + i] = getScaledDims(
rewriter, loc, dims[hDimOffset + i], scaleFactorsFloatValues[i]);
} else {
SmallVector<Value, 2> outputSizeTorchInt;
if (!getListConstructElements(op.output_size(), outputSizeTorchInt))
return rewriter.notifyMatchFailure(
op, "unimplemented: the output_size is not constructed from "
"ListConstruct");
SmallVector<Value, 2> outputSizeIntValues;
outputSizeIntValues = getTypeConvertedValues(
rewriter, loc, getTypeConverter(), outputSizeTorchInt);
for (unsigned i = 0; i < outputSizeTorchInt.size(); i++) {
auto scaleFactorVal = getScaleFactor(
rewriter, loc, dims[hDimOffset + i], outputSizeIntValues[i]);
scaleFactorsInt.push_back(scaleFactorVal);
dims[hDimOffset + i] =
castIntToIndex(rewriter, loc, outputSizeIntValues[i]);
}
}
Value outTensor = rewriter.create<tensor::EmptyOp>(
loc, getAsOpFoldResult(dims), elementType);
AffineMap idMap = rewriter.getMultiDimIdentityMap(inputRank);
SmallVector<StringRef> iteratorTypes(inputRank,
getParallelIteratorTypeName());
Value finalRes =
rewriter
.create<linalg::GenericOp>(
loc, outTensor.getType(), ValueRange{}, outTensor,
/*indexingMaps=*/idMap,
/*iteratorTypes=*/iteratorTypes,
[&](OpBuilder &b, Location loc, ValueRange args) {
SmallVector<Value> indices;
for (unsigned i = 0; i < inputRank; i++)
indices.push_back(b.create<linalg::IndexOp>(loc, i));
for (unsigned i = 0; i < (inputRank - hDimOffset); i++)
indices[i + hDimOffset] = b.create<arith::FloorDivSIOp>(
loc, indices[i + hDimOffset],
castIntToIndex(rewriter, loc, scaleFactorsInt[i]));
Value retVal =
b.create<tensor::ExtractOp>(loc, input, indices);
b.create<linalg::YieldOp>(loc, retVal);
})
.getResult(0);
rewriter.replaceOpWithNewOp<tensor::CastOp>(op, resultType, finalRes);
return success();
}
};
} // namespace
static Value getGradOutputValue(OpBuilder &builder, Location loc,
Value gradOutput, Type gradOutputElemType,
Value numBatch, Value numChannel,
Value inputIndexH, Value inputIndexW,
Value kernelIndexH, Value kernelIndexW,
SmallVector<Value> &gradOutputSizeIndexValues,
SmallVector<Value, 2> &scaleFactorsIntValues) {
Value constantOne = builder.create<arith::ConstantIndexOp>(loc, 1);
Value outputIndexH = builder.create<arith::MulIOp>(
loc, inputIndexH, castIntToIndex(builder, loc, scaleFactorsIntValues[0]));
outputIndexH = builder.create<arith::AddIOp>(loc, outputIndexH, kernelIndexH);
Value outputIndexW = builder.create<arith::MulIOp>(
loc, inputIndexW, castIntToIndex(builder, loc, scaleFactorsIntValues[1]));
outputIndexW = builder.create<arith::AddIOp>(loc, outputIndexW, kernelIndexW);
// Handling corner cases.
Value gradOutputHMinusOne = builder.create<arith::SubIOp>(
loc, gradOutputSizeIndexValues[2], constantOne);
Value predH = builder.create<arith::CmpIOp>(
loc, arith::CmpIPredicate::sle, outputIndexH, gradOutputHMinusOne);
outputIndexH = builder.create<arith::SelectOp>(loc, predH, outputIndexH,
gradOutputHMinusOne);
Value gradOutputWMinusOne = builder.create<arith::SubIOp>(
loc, gradOutputSizeIndexValues[3], constantOne);
Value predW = builder.create<arith::CmpIOp>(
loc, arith::CmpIPredicate::sle, outputIndexW, gradOutputWMinusOne);
outputIndexW = builder.create<arith::SelectOp>(loc, predW, outputIndexW,
gradOutputWMinusOne);
Value gradOutputValue = builder.create<tensor::ExtractOp>(
loc, gradOutput,
ValueRange{numBatch, numChannel, outputIndexH, outputIndexW});
Value constantZero =
builder.create<arith::ConstantOp>(loc, builder.getF32FloatAttr(0.0));
Value pred = builder.create<arith::AndIOp>(loc, predH, predW);
Value result = builder.create<arith::SelectOp>(
loc, pred, gradOutputValue,
convertScalarToDtype(builder, loc, constantZero, gradOutputElemType));
return result;
}
// The implementation of the `aten.upsample_nearest2d_backward.vec` op's
// lowering is as follows:
// gradOutput: Tensor of size [n, c, oh, ow]
// outTensor: Tensor of size [n, c, ih, iw], initialized with zero
// kh = ceil(oh/ih), kw = ceil(ow/iw)
//
// for i in range(n):
// for j in range(c):
// for p in range(ih):
// for q in range(iw):
// for x in range(kh):
// for y in range(kw):
// outTensor[i, j, p, q] += gradOutput[i, j, (p*kh)+x, (q*kw)+y]
namespace {
class ConvertAtenUpsampleNearest2dBackwardVecOp
: public OpConversionPattern<AtenUpsampleNearest2dBackwardVecOp> {
public:
using OpConversionPattern::OpConversionPattern;
LogicalResult
matchAndRewrite(AtenUpsampleNearest2dBackwardVecOp op, OpAdaptor adaptor,
ConversionPatternRewriter &rewriter) const override {
Location loc = op->getLoc();
Value gradOutput = adaptor.grad_output();
Type resultType = getTypeConverter()->convertType(op.getResult().getType());
auto gradOutputType = gradOutput.getType().cast<RankedTensorType>();
auto gradOutputRank = gradOutputType.getRank();
Type elementType = gradOutputType.getElementType();
SmallVector<Value> gradOutputSizeIndexValues =
getTensorSizes(rewriter, loc, gradOutput);
SmallVector<Value> gradOutputSizeIntValues =
castIndexVectorToInt64Vector(rewriter, loc, gradOutputSizeIndexValues);
SmallVector<Value, 2> scaleFactorsFloatValues;
SmallVector<Value, 4> inputSizeTorchInt;
if (!getListConstructElements(op.input_size(), inputSizeTorchInt))
return rewriter.notifyMatchFailure(
op, "unimplemented: the input_size is not constructed from "
"ListConstruct");
SmallVector<Value, 4> inputSizeIntValues;
inputSizeIntValues = getTypeConvertedValues(
rewriter, loc, getTypeConverter(), inputSizeTorchInt);
// The dimension at which the scaling starts.
unsigned hDimOffset = 2;
if (!op.scale_factors().getType().isa<Torch::NoneType>()) {
SmallVector<Value, 2> scaleFactorsTorchFloat;
if (!getListConstructElements(op.scale_factors(), scaleFactorsTorchFloat))
return rewriter.notifyMatchFailure(
op, "unimplemented: the scale_factors is not constructed from "
"ListConstruct");
scaleFactorsFloatValues = getTypeConvertedValues(
rewriter, loc, getTypeConverter(), scaleFactorsTorchFloat);
} else {
for (unsigned i = hDimOffset; i < gradOutputRank; i++) {
auto scaleFactorVal = rewriter.create<arith::DivFOp>(
loc,
convertScalarToDtype(rewriter, loc, gradOutputSizeIntValues[i],
mlir::Float32Type::get(op->getContext())),
convertScalarToDtype(rewriter, loc, inputSizeIntValues[i],
mlir::Float32Type::get(op->getContext())));
scaleFactorsFloatValues.push_back(scaleFactorVal);
}
}
SmallVector<Value, 2> scaleFactorsIntValues;
for (auto v : scaleFactorsFloatValues)
scaleFactorsIntValues.push_back(convertScalarToDtype(
rewriter, loc, rewriter.create<math::CeilOp>(loc, v),
mlir::IntegerType::get(op->getContext(), 64)));
Value outTensor = createZeroInitTensor(
rewriter, loc,
castIntVectorToIndexVector(rewriter, loc, inputSizeIntValues),
elementType);
Value kernelTensor = rewriter.create<tensor::EmptyOp>(
loc,
getAsOpFoldResult(
castIntVectorToIndexVector(rewriter, loc, scaleFactorsIntValues)),
elementType);
unsigned kernelRank = scaleFactorsIntValues.size();
SmallVector<AffineExpr> affineExprs;
for (unsigned i = 0; i < gradOutputRank; i++)
affineExprs.push_back(rewriter.getAffineDimExpr(i));
AffineMap outputMap =
AffineMap::get(gradOutputRank + kernelRank,
/*symbolCount=*/0, affineExprs, op->getContext());
affineExprs.clear();
for (unsigned i = gradOutputRank; i < gradOutputRank + kernelRank; i++)
affineExprs.push_back(rewriter.getAffineDimExpr(i));
AffineMap kernelMap =
AffineMap::get(gradOutputRank + kernelRank,
/*symbolCount=*/0, affineExprs, op->getContext());
SmallVector<AffineMap> indexingMaps{kernelMap, outputMap};
SmallVector<StringRef> iteratorTypes(gradOutputRank,
getParallelIteratorTypeName());
iteratorTypes.push_back(getReductionIteratorTypeName());
iteratorTypes.push_back(getReductionIteratorTypeName());
Value finalRes =
rewriter
.create<linalg::GenericOp>(
loc, outTensor.getType(), ValueRange{kernelTensor},
ValueRange{outTensor},
/*indexingMaps=*/indexingMaps,
/*iteratorTypes=*/iteratorTypes,
[&](OpBuilder &b, Location loc, ValueRange args) {
Value n = rewriter.create<linalg::IndexOp>(loc, 0);
Value c = rewriter.create<linalg::IndexOp>(loc, 1);
Value ih = rewriter.create<linalg::IndexOp>(loc, 2);
Value iw = rewriter.create<linalg::IndexOp>(loc, 3);
Value kh = rewriter.create<linalg::IndexOp>(loc, 4);
Value kw = rewriter.create<linalg::IndexOp>(loc, 5);
Value accValue = getGradOutputValue(
rewriter, loc, gradOutput, elementType, n, c, ih, iw, kh,
kw, gradOutputSizeIndexValues, scaleFactorsIntValues);
Value outputVal = args[1];
outputVal =
rewriter.create<arith::AddFOp>(loc, outputVal, accValue);
b.create<linalg::YieldOp>(loc, outputVal);
})
->getResult(0);
rewriter.replaceOpWithNewOp<tensor::CastOp>(op, resultType, finalRes);
return success();
}
};
} // namespace
void mlir::torch::torch_to_linalg::
populateIndirectDataMovementPatternsAndLegality(
TypeConverter &typeConverter, RewritePatternSet &patterns,
ConversionTarget &target) {
MLIRContext *context = patterns.getContext();
target.addIllegalOp<AtenGatherOp>();
patterns.add<ConvertAtenGatherOp>(typeConverter, context);
target.addIllegalOp<AtenEmbeddingOp>();
patterns.add<ConvertAtenEmbeddingOp>(typeConverter, context);
target.addIllegalOp<AtenIndexSelectOp>();
patterns.add<ConvertAtenIndexSelectOp>(typeConverter, context);
target.addIllegalOp<AtenIndexTensorOp>();
patterns.add<ConvertAtenIndexTensorOp>(typeConverter, context);
target.addIllegalOp<AtenEmbeddingBagPaddingIdxOp>();
patterns.add<ConvertAtenEmbeddingBagPaddingIdxOp>(typeConverter, context);
target.addIllegalOp<AtenUpsampleNearest2dVecOp>();
patterns.add<ConvertAtenUpsampleNearest2dVecOp>(typeConverter, context);
target.addIllegalOp<AtenUpsampleNearest2dBackwardVecOp>();
patterns.add<ConvertAtenUpsampleNearest2dBackwardVecOp>(typeConverter, context);
}