shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
llvm-project/mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorRewriting.cpp
Aart Bik 80fe3168b5 [mlir][sparse] add support for direct prod/and/min/max reductions 
We recently fixed a bug in "sparsifying" such reductions, since
it incorrectly changed this into reductions over stored elements
only , which only works for add/sub/or/xor. However, we still want
to be able to "sparsify" the reductions even in the general case,
and this is a first step by rewriting them into a custom reduction
that feeds in the implicit zeros. NOTE HOWEVER, that in the long run
we want to do this better and feed in any implicit zero only ONCE
for efficiency.

Reviewed By: Peiming

Differential Revision: https://reviews.llvm.org/D152580
2023-06-12 09:27:47 -07:00

1372 lines
57 KiB
C++
Raw Blame History

//===- SparseTensorRewriting.cpp - Sparse tensor rewriting rules ----------===//
//
// 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
//
//===----------------------------------------------------------------------===//
//
// This file implements rewriting rules that are specific to sparse tensors.
//
//===----------------------------------------------------------------------===//
#include "CodegenUtils.h"
#include "LoopEmitter.h"
#include "mlir/Dialect/Arith/IR/Arith.h"
#include "mlir/Dialect/Bufferization/IR/Bufferization.h"
#include "mlir/Dialect/Linalg/IR/Linalg.h"
#include "mlir/Dialect/Linalg/Utils/Utils.h"
#include "mlir/Dialect/MemRef/IR/MemRef.h"
#include "mlir/Dialect/SCF/IR/SCF.h"
#include "mlir/Dialect/SparseTensor/IR/SparseTensor.h"
#include "mlir/Dialect/SparseTensor/IR/SparseTensorType.h"
#include "mlir/Dialect/SparseTensor/Transforms/Passes.h"
#include "mlir/Dialect/Tensor/IR/Tensor.h"
#include "mlir/IR/AffineMap.h"
#include "mlir/IR/Matchers.h"
#include "mlir/Support/LLVM.h"
using namespace mlir;
using namespace mlir::bufferization;
using namespace mlir::linalg;
using namespace mlir::sparse_tensor;
//===---------------------------------------------------------------------===//
// Helper methods for the actual rewriting rules.
//===---------------------------------------------------------------------===//
// Helper method to match any typed zero.
static bool isZeroValue(Value val) {
return matchPattern(val, m_Zero()) || matchPattern(val, m_AnyZeroFloat());
}
// Helper to detect a sparse tensor type operand.
static bool isSparseTensor(OpOperand *op) {
auto enc = getSparseTensorEncoding(op->get().getType());
return enc && llvm::is_contained(enc.getLvlTypes(), DimLevelType::Compressed);
}
// Helper method to find zero/uninitialized allocation.
static bool isAlloc(OpOperand *op, bool isZero) {
Value val = op->get();
// Check allocation, with zero alloc when required.
if (auto alloc = val.getDefiningOp<AllocTensorOp>()) {
Value copy = alloc.getCopy();
if (isZero)
return copy && isZeroValue(copy);
return !copy;
}
// Last resort for zero alloc: the whole value is zero.
return isZero && isZeroValue(val);
}
// Helper to detect sampling operation.
static bool isSampling(GenericOp op) {
auto yieldOp = cast<linalg::YieldOp>(op.getRegion().front().getTerminator());
if (auto *def = yieldOp.getOperand(0).getDefiningOp()) {
if (isa<arith::MulFOp>(def) || isa<arith::MulIOp>(def)) {
// Both scalar input arguments used exactly once.
Value s1 = op.getBlock()->getArgument(0);
Value s2 = op.getBlock()->getArgument(1);
return (def->getOperand(0) == s1 && def->getOperand(1) == s2) ||
(def->getOperand(1) == s1 && def->getOperand(0) == s2);
}
}
return false;
}
// Helper to detect chain of multiplications that do not involve x.
static bool isMulChain(Value val, Value x) {
if (auto arg = dyn_cast<BlockArgument>(val))
return arg != x;
if (auto *def = val.getDefiningOp()) {
if (isa<arith::MulFOp>(def) || isa<arith::MulIOp>(def))
return isMulChain(def->getOperand(0), x) &&
isMulChain(def->getOperand(1), x);
}
return false;
}
// Helper to detect x = x + <multiplications>.
static bool isSumOfMul(GenericOp op) {
auto yieldOp = cast<linalg::YieldOp>(op.getRegion().front().getTerminator());
if (auto *def = yieldOp.getOperand(0).getDefiningOp()) {
if (isa<arith::AddFOp>(def) || isa<arith::AddIOp>(def)) {
Value x = op.getBlock()->getArguments().back();
return (def->getOperand(0) == x && isMulChain(def->getOperand(1), x)) ||
(def->getOperand(1) == x && isMulChain(def->getOperand(0), x));
}
}
return false;
}
// Helper to detect direct yield of a zero value.
static bool isZeroYield(GenericOp op) {
auto yieldOp = cast<linalg::YieldOp>(op.getRegion().front().getTerminator());
if (auto arg = dyn_cast<BlockArgument>(yieldOp.getOperand(0))) {
if (arg.getOwner()->getParentOp() == op) {
return isZeroValue(op->getOperand(arg.getArgNumber()));
}
}
return isZeroValue(yieldOp.getOperand(0));
}
/// Populates given sizes array from type (for static sizes) and from
/// the tensor (for dynamic sizes).
static void sizesForTensor(OpBuilder &builder, SmallVectorImpl<Value> &sizes,
Location loc, ShapedType stp, Value tensor) {
for (const auto &d : enumerate(stp.getShape())) {
Value dim;
if (d.value() == ShapedType::kDynamic)
dim = builder.create<tensor::DimOp>(loc, tensor, d.index());
else
dim = constantIndex(builder, loc, d.value());
sizes.push_back(dim);
}
}
// TODO: The dim level property of the COO type relies on input tensors, the
// shape relies on the output tensor
static RankedTensorType getCOOType(const SparseTensorType &stt, bool ordered) {
return getCOOFromTypeWithOrdering(stt, stt.getDimToLvl(), ordered);
}
static RankedTensorType getBufferType(const SparseTensorType &stt,
bool needTmpCOO) {
return needTmpCOO ? getCOOType(stt, /*ordered=*/false)
: stt.getRankedTensorType();
}
/// Collects the dynamic dimension sizes for `tp` with the assumption that
/// `sizes` are the dimension sizes for the type. Stores the dynamic dimension
/// sizes to dynSizes.
static void getDynamicSizes(RankedTensorType tp,
const SmallVectorImpl<Value> &sizes,
SmallVectorImpl<Value> &dynSizes) {
for (const auto &d : enumerate(tp.getShape())) {
if (d.value() == ShapedType::kDynamic)
dynSizes.push_back(sizes[d.index()]);
}
}
static LogicalResult genForeachOnSparseConstant(ForeachOp op,
RewriterBase &rewriter,
SparseElementsAttr attr) {
auto loc = op.getLoc();
SmallVector<Value> reduc = op.getInitArgs();
// Foreach on constant.
foreachInSparseConstant(
rewriter, loc, attr, op.getOrder().value_or(AffineMap()),
[&reduc, &rewriter, op](ArrayRef<Value> cvs, Value v) mutable {
SmallVector<Value> args;
args.append(cvs.begin(), cvs.end());
args.push_back(v);
args.append(reduc);
// Clones the foreach op to get a copy of the loop body.
auto cloned = cast<ForeachOp>(rewriter.clone(*op.getOperation()));
assert(args.size() == cloned.getBody()->getNumArguments());
Operation *yield = cloned.getBody()->getTerminator();
rewriter.inlineBlockBefore(cloned.getBody(), op, args);
// clean up
rewriter.eraseOp(cloned);
reduc = yield->getOperands();
rewriter.eraseOp(yield);
});
rewriter.replaceOp(op, reduc);
return success();
}
/// Populates the given sizes array for concatenation from types (for static
/// sizes) and from the source tensors (for dynamic sizes).
static void concatSizesFromInputs(OpBuilder &builder,
SmallVectorImpl<Value> &sizes, Location loc,
ShapedType dstTp, ValueRange srcs,
unsigned dim) {
auto dstShape = dstTp.getShape();
sizesFromSrc(builder, sizes, loc, srcs[0]);
// Sum up on the `dim` if the dimension is dynamic.
if (dstShape[dim] != ShapedType::kDynamic) {
// Faithfully take the static size.
sizes[dim] = constantIndex(builder, loc, dstShape[dim]);
} else {
// Else, compute the shape dynamically.
for (const auto &src : srcs.drop_front()) {
Value srcSz = linalg::createOrFoldDimOp(builder, loc, src, dim);
// Sum up all the sizes.
sizes[dim] = builder.create<arith::AddIOp>(loc, sizes[dim], srcSz);
}
}
}
//===---------------------------------------------------------------------===//
// The actual sparse tensor rewriting rules.
//===---------------------------------------------------------------------===//
namespace {
/// Rewriting rule that converts direct yield of zero with initial allocation.
struct FoldInvariantYield : public OpRewritePattern<GenericOp> {
public:
using OpRewritePattern<GenericOp>::OpRewritePattern;
LogicalResult matchAndRewrite(GenericOp op,
PatternRewriter &rewriter) const override {
if (!op.hasTensorSemantics() || op.getNumResults() != 1 ||
!isAlloc(op.getDpsInitOperand(0), /*isZero=*/false) ||
!isZeroYield(op) || !op.getDpsInitOperand(0)->get().hasOneUse())
return failure();
auto outputType = getRankedTensorType(op.getResult(0));
// Yielding zero on newly allocated (all-zero) sparse tensors can be
// optimized out directly (regardless of dynamic or static size).
if (getSparseTensorEncoding(outputType)) {
rewriter.replaceOp(op, op.getDpsInitOperand(0)->get());
return success();
}
// Incorporate zero value into allocation copy.
if (!outputType.hasStaticShape())
return failure();
Value zero = constantZero(rewriter, op.getLoc(), op.getResult(0).getType());
AllocTensorOp a =
op.getDpsInitOperand(0)->get().getDefiningOp<AllocTensorOp>();
rewriter.updateRootInPlace(a, [&]() { a.getCopyMutable().assign(zero); });
rewriter.replaceOp(op, op.getDpsInitOperand(0)->get());
return success();
}
};
/// Rewriting rule that converts two kernels:
///
/// T(i,j) = SUM(k, A(i,j,k) * B(i,j,k) * ... )
/// X(i,j) = S(i,j) * T(i,j)
///
/// into a single kernel, using distributive law:
///
/// X(i,j) = SUM(k, S(i,j) * A(i,j,k) * B(i,j,k) * ... )
///
/// This kind of fusion (merging two ops into one but using arithmetic
/// equalities that may not hold for floating-point computations) would
/// be undesirable in the dense case, since we distribute the multiplication
/// into the reduction loop. However, for sparse sampling tensor S, such
/// a fusion may actually reduce the asymptotic complexity of the kernel,
/// since intermediate results may be nullified.
struct FuseSparseMultiplyOverAdd : public OpRewritePattern<GenericOp> {
public:
using OpRewritePattern<GenericOp>::OpRewritePattern;
LogicalResult matchAndRewrite(GenericOp op,
PatternRewriter &rewriter) const override {
// Check consumer.
if (!op.hasTensorSemantics() || op.getNumDpsInputs() != 2 ||
op.getNumResults() != 1 ||
op.getNumParallelLoops() != op.getNumLoops() ||
!op.getMatchingIndexingMap(op.getDpsInitOperand(0)).isIdentity() ||
!op.getMatchingIndexingMap(op.getDpsInputOperand(0)).isIdentity() ||
!op.getMatchingIndexingMap(op.getDpsInputOperand(1)).isIdentity())
return failure();
// Find consuming OP2(sparse, other) or OP2(other, sparse). The other
// operand can be sparse or dense, since the point of this rewriting rule
// is detecting a situation in which *more* sparsity is introduced into
// a computation, be it already sparse or still dense.
unsigned other = 0;
if (isSparseTensor(op.getDpsInputOperand(0)))
other = 1;
else if (!isSparseTensor(op.getDpsInputOperand(1)))
return failure();
// Check producer.
auto prod = dyn_cast_or_null<GenericOp>(
op.getDpsInputOperand(other)->get().getDefiningOp());
if (!prod || !prod.hasTensorSemantics() || prod.getNumResults() != 1 ||
!prod.getResult(0).hasOneUse())
return failure();
// Sampling consumer and sum of multiplication chain producer.
if (!isAlloc(op.getDpsInitOperand(0), /*isZero=*/false) ||
!isAlloc(prod.getDpsInitOperand(0), /*isZero=*/true) ||
!isSampling(op) || !isSumOfMul(prod))
return failure();
// Modify operand structure of producer and consumer.
Location loc = prod.getLoc();
SmallVector<Value> inputOps = prod.getInputs();
SmallVector<Value> outputOps = op.getOutputs();
SmallVector<AffineMap> fusedIndexMaps = prod.getIndexingMapsArray();
inputOps.push_back(op.getDpsInputOperand(1 - other)->get());
fusedIndexMaps.push_back(fusedIndexMaps.back()); // mimic other
// Fuse producer and consumer into a new generic op.
auto fusedOp = rewriter.create<GenericOp>(
loc, op.getResult(0).getType(), inputOps, outputOps,
rewriter.getAffineMapArrayAttr(fusedIndexMaps), prod.getIteratorTypes(),
/*doc=*/nullptr, /*library_call=*/nullptr);
Block &prodBlock = prod.getRegion().front();
Block &consBlock = op.getRegion().front();
IRMapping mapper;
Block *fusedBlock = new Block();
fusedOp.getRegion().push_back(fusedBlock);
unsigned num = prodBlock.getNumArguments();
for (unsigned i = 0; i < num - 1; i++)
addArg(mapper, fusedBlock, prodBlock.getArgument(i));
addArg(mapper, fusedBlock, consBlock.getArgument(1 - other));
addArg(mapper, fusedBlock, prodBlock.getArgument(num - 1));
// Clone bodies of the producer and consumer in new evaluation order.
auto *acc = prodBlock.getTerminator()->getOperand(0).getDefiningOp();
auto *sampler = consBlock.getTerminator()->getOperand(0).getDefiningOp();
rewriter.setInsertionPointToStart(fusedBlock);
Value last;
for (auto &op : prodBlock.without_terminator())
if (&op != acc) {
last = op.getResult(0);
rewriter.clone(op, mapper);
}
mapper.map(consBlock.getArgument(other), fusedBlock->back().getResult(0));
mapper.map(last, rewriter.clone(*sampler, mapper)->getResult(0));
last = rewriter.clone(*acc, mapper)->getResult(0);
rewriter.create<linalg::YieldOp>(loc, last);
// Force initial value on merged allocation for dense outputs.
if (!getSparseTensorEncoding(op.getResult(0).getType())) {
Value init = prod.getDpsInitOperand(0)
->get()
.getDefiningOp<AllocTensorOp>()
.getCopy();
AllocTensorOp a =
op.getDpsInitOperand(0)->get().getDefiningOp<AllocTensorOp>();
rewriter.updateRootInPlace(a, [&]() { a.getCopyMutable().assign(init); });
}
// Replace consumer with fused operation. Old producer
// and consumer ops will be removed by DCE.
rewriter.replaceOp(op, fusedOp->getResults());
return success();
}
private:
// Helper to add argument and record the mapping.
static void addArg(IRMapping &mapper, Block *b, BlockArgument a) {
mapper.map(a, b->addArgument(a.getType(), a.getLoc()));
}
};
// Fuse a tensor cast into producing operation. Note that a tensor.cast
// should really not be used to convert between sparse encodings. Since
// the pattern currently appears as a result of some prior rewriting
// we make an attempt to repair very obvious cases.
// TODO: audit the pure tensor dialect rewriting rules
struct FuseTensorCast : public OpRewritePattern<tensor::CastOp> {
public:
using OpRewritePattern<tensor::CastOp>::OpRewritePattern;
LogicalResult matchAndRewrite(tensor::CastOp op,
PatternRewriter &rewriter) const override {
Type srcType = op.getSource().getType();
Type dstType = op.getDest().getType();
// A nop cast simply folds away.
if (srcType == dstType) {
rewriter.replaceOp(op, op->getResults());
return success();
}
// See if a sparsity changing cast can be fused into producer.
if (tensor::isSameTypeWithoutEncoding(srcType, dstType)) {
if (Operation *def = op.getSource().getDefiningOp()) {
if (def->hasOneUse() && isa<tensor::ExtractSliceOp>(def)) {
rewriter.updateRootInPlace(def, [&]() {
def->getResult(0).setType(op->getResultTypes()[0]);
});
rewriter.replaceOp(op, def->getResult(0));
return success();
}
}
}
// Repair tensor casts with at least one sparse operand into the
// the properly supported sparse_tensor.convert.
if (getSparseTensorEncoding(srcType) || getSparseTensorEncoding(dstType)) {
rewriter.replaceOpWithNewOp<ConvertOp>(op, dstType, op.getSource());
return success();
}
// Fail otherwise.
return failure();
}
};
/// Rewrites a sparse reduction that would not sparsify directly since
/// doing so would only iterate over the stored elements, ignoring the
/// implicit zeros, into a semi-ring. Applies to all prod/and/min/max
/// (note that reductions like add/sub/or/xor can directly be sparsified
/// since the implicit zeros do not contribute to the final result).
/// Note that prod/and are still included since, even though they often
/// are nullified in sparse data, they may still occur for special
/// situations in which e.g. some rows in a sparse matrix are fully
/// dense. For min/max, including the implicit zeros is a much more
/// common situation.
///
/// TODO: this essentially "densifies" the operation; we want to implement
/// this much more efficiently by performing the reduction over the
/// stored values, and feed in the zero once if there were *any*
/// implicit zeros as well; but for now, at least we provide
/// the functionality
///
struct GenSemiRingReduction : public OpRewritePattern<GenericOp> {
public:
using OpRewritePattern<GenericOp>::OpRewritePattern;
LogicalResult matchAndRewrite(GenericOp op,
PatternRewriter &rewriter) const override {
// Reject non-reductions.
if (!op.hasTensorSemantics() || op.getNumDpsInputs() != 1 ||
op.getNumReductionLoops() == 0 || op.getNumResults() != 1)
return failure();
auto inp = op.getDpsInputOperand(0);
auto init = op.getDpsInitOperand(0);
if (!isSparseTensor(inp))
return failure();
// Look for direct x = x OP y for semi-ring ready reductions.
auto red = cast<linalg::YieldOp>(op.getRegion().front().getTerminator())
.getOperand(0)
.getDefiningOp();
if (!isa<arith::AndIOp, arith::MulIOp, arith::MulFOp, arith::MinFOp,
arith::MinSIOp, arith::MinUIOp, arith::MaxFOp, arith::MaxSIOp,
arith::MaxUIOp>(red))
return failure();
Value s0 = op.getBlock()->getArgument(0);
Value s1 = op.getBlock()->getArgument(1);
if ((red->getOperand(0) != s0 || red->getOperand(1) != s1) &&
(red->getOperand(0) != s1 || red->getOperand(1) != s0))
return failure();
// Identity.
Location loc = op.getLoc();
Value identity =
rewriter.create<tensor::ExtractOp>(loc, init->get(), ValueRange());
// Unary {
// present -> value
// absent -> zero.
// }
Type rtp = s0.getType();
rewriter.setInsertionPointToStart(&op.getRegion().front());
auto semiring = rewriter.create<sparse_tensor::UnaryOp>(loc, rtp, s0);
Block *present =
rewriter.createBlock(&semiring.getPresentRegion(), {}, rtp, loc);
rewriter.setInsertionPointToStart(&semiring.getPresentRegion().front());
rewriter.create<sparse_tensor::YieldOp>(loc, present->getArgument(0));
rewriter.createBlock(&semiring.getAbsentRegion(), {}, {}, {});
rewriter.setInsertionPointToStart(&semiring.getAbsentRegion().front());
auto zero =
rewriter.create<arith::ConstantOp>(loc, rewriter.getZeroAttr(rtp));
rewriter.create<sparse_tensor::YieldOp>(loc, zero);
rewriter.setInsertionPointAfter(semiring);
// CustomReduce {
// x = x REDUC y, identity
// }
auto custom = rewriter.create<sparse_tensor::ReduceOp>(
loc, rtp, semiring.getResult(), s1, identity);
Block *region =
rewriter.createBlock(&custom.getRegion(), {}, {rtp, rtp}, {loc, loc});
rewriter.setInsertionPointToStart(&custom.getRegion().front());
IRMapping irMap;
irMap.map(red->getOperand(0), region->getArgument(0));
irMap.map(red->getOperand(1), region->getArgument(1));
auto cloned = rewriter.clone(*red, irMap);
rewriter.create<sparse_tensor::YieldOp>(loc, cloned->getResult(0));
rewriter.setInsertionPointAfter(custom);
rewriter.replaceOp(red, custom.getResult());
return success();
}
};
/// Sparse rewriting rule for sparse-to-sparse reshape operator.
struct TensorReshapeRewriter : public OpRewritePattern<tensor::ReshapeOp> {
public:
using OpRewritePattern<tensor::ReshapeOp>::OpRewritePattern;
LogicalResult matchAndRewrite(tensor::ReshapeOp op,
PatternRewriter &rewriter) const override {
Location loc = op.getLoc();
Value srcTensor = op.getSource();
const auto srcTp = getSparseTensorType(srcTensor);
const auto dstTp = getSparseTensorType(op.getResult());
if (!srcTp.hasEncoding() || !dstTp.hasEncoding() ||
!dstTp.hasStaticDimShape())
return failure();
SmallVector<Value> srcSizes;
sizesForTensor(rewriter, srcSizes, loc, srcTp, srcTensor);
SmallVector<Value> dstSizes;
for (Dimension d : dstTp.getDimShape())
dstSizes.push_back(constantIndex(rewriter, loc, d));
Value nnz = rewriter.create<NumberOfEntriesOp>(loc, srcTensor);
// Only need an unordered COO buffer if input and output are not sorted
// in the same way.
Type bufferTp = getBufferType(
dstTp.withoutDimToLvl(),
!srcTp.isAllOrdered() || !srcTp.isIdentity() || !dstTp.isIdentity());
SmallVector<Value> dynSizes;
Value buffer = rewriter
.create<AllocTensorOp>(loc, bufferTp, dynSizes, Value(),
nnz, Attribute())
.getResult();
// Convert src coordinates to dst coordinates by first collapsing it to 1D
// and then expand it to the match the rank of the destination tensor.
// Implemented as follows:
// foreach srcCoords %srcTensor
// collapsedCoords = reshapeCvs(srcCoords, [1, ..., srcRank])
// expandedCoords = reshapeCvs(collapsedCoords, [1, ..., dstRank])
// insert expandedCoords, %buffer
//
// followed by an optional
// %t = sparse_tensor.cast %tmp
// depending on whether the input/output are sorted in the same way.
const auto encSrc = srcTp.getEncoding();
ForeachOp foreachOp = rewriter.create<ForeachOp>(
loc, srcTensor, buffer,
[&](OpBuilder &builder, Location loc, ValueRange srcLcvs, Value v,
ValueRange reduc) {
const Dimension srcRank = srcTp.getDimRank();
SmallVector<Value> srcDcvs;
srcDcvs.reserve(srcRank);
for (Dimension d = 0; d < srcRank; d++) {
// FIXME: `toStoredDim` is deprecated
Level lvl = toStoredDim(encSrc, d);
srcDcvs.push_back(srcLcvs[lvl]);
}
Value collapsed_size = constantIndex(builder, loc, 1);
for (Dimension d = 0; d < srcRank; d++)
collapsed_size =
builder.create<arith::MulIOp>(loc, collapsed_size, srcSizes[d]);
SmallVector<Value, 1> collapsedSizes = {collapsed_size};
ReassociationIndices collapse_indices;
for (Dimension i = 0; i < srcRank; i++)
collapse_indices.push_back(i);
SmallVector<ReassociationIndices, 1> collapse_reassociation = {
collapse_indices};
SmallVector<Value, 1> collapsedDcvs;
reshapeCvs(builder, loc, collapse_reassociation, srcSizes, srcDcvs,
collapsedSizes, collapsedDcvs);
ReassociationIndices expand_indices;
for (Dimension i = 0; i < dstTp.getDimRank(); i++)
expand_indices.push_back(i);
SmallVector<ReassociationIndices, 1> expand_reassociation = {
expand_indices};
SmallVector<Value> dstDcvs;
reshapeCvs(builder, loc, expand_reassociation, collapsedSizes,
collapsedDcvs, dstSizes, dstDcvs);
auto t = builder.create<InsertOp>(loc, v, reduc.front(), dstDcvs);
builder.create<sparse_tensor::YieldOp>(loc, t);
});
Value t = rewriter.create<LoadOp>(loc, foreachOp.getResult(0), true);
if (bufferTp != dstTp) {
auto dstRTT = dstTp.getRankedTensorType();
Value converted = rewriter.create<ConvertOp>(loc, dstRTT, t).getResult();
rewriter.create<DeallocTensorOp>(loc, t);
t = converted;
}
rewriter.replaceOp(op, t);
return success();
}
};
/// Sparse rewriting rule for sparse-to-sparse reshape operator.
template <typename ReshapeOp>
struct Sparse2SparseReshapeRewriter : public OpRewritePattern<ReshapeOp> {
public:
using OpRewritePattern<ReshapeOp>::OpRewritePattern;
LogicalResult matchAndRewrite(ReshapeOp op,
PatternRewriter &rewriter) const override {
Location loc = op.getLoc();
Value srcTensor = op.getSrc();
const auto srcTp = getSparseTensorType(srcTensor);
const auto dstTp = getSparseTensorType(op.getResult());
if (!srcTp.hasEncoding() || !dstTp.hasEncoding())
return failure();
// Generate code to represent the static dimension constants or compute
// the dynamic dimension values.
SmallVector<Value> srcSizes;
sizesForTensor(rewriter, srcSizes, loc, srcTp, srcTensor);
SmallVector<Value> dstSizes;
SmallVector<Value> dstDynSizes;
if (dstTp.hasStaticDimShape()) {
for (Dimension d : dstTp.getDimShape())
dstSizes.push_back(constantIndex(rewriter, loc, d));
} else {
ArrayRef<DynSize> dstShape = dstTp.getDimShape();
genReshapeDstShape(rewriter, loc, dstSizes, srcSizes, dstShape,
op.getReassociationIndices());
for (auto [idx, shape] : llvm::enumerate(dstShape)) {
if (shape == ShapedType::kDynamic)
dstDynSizes.push_back(dstSizes[idx]);
}
}
Value nnz = rewriter.create<NumberOfEntriesOp>(loc, srcTensor);
// Only need a unordered COO buffer if input and output are not sorted
// in the same way.
Type bufferTp = getBufferType(
dstTp.withoutDimToLvl(),
!srcTp.isAllOrdered() || !srcTp.isIdentity() || !dstTp.isIdentity());
Value buffer =
rewriter
.create<AllocTensorOp>(loc, bufferTp, dstDynSizes, Value(),
/*sizeHint=*/nnz, Attribute())
.getResult();
// Implement the sparse2sparse reshape as follows:
// foreach srcCoords %srcTensor
// insert reshapeCvs(srcCoords), %buffer
//
// followed by an optional
// %t = sparse_tensor.cast %tmp
// depending on whether the input/output are sorted in the same way.
const auto encSrc = srcTp.getEncoding();
ForeachOp foreachOp = rewriter.create<ForeachOp>(
loc, srcTensor, buffer,
[&](OpBuilder &builder, Location loc, ValueRange srcLcvs, Value v,
ValueRange reduc) {
const Dimension dimRank = srcTp.getDimRank();
SmallVector<Value> srcDcvs;
srcDcvs.reserve(dimRank);
for (Dimension d = 0; d < dimRank; d++) {
// FIXME: `toStoredDim` is deprecated
Level lvl = toStoredDim(encSrc, d);
srcDcvs.push_back(srcLcvs[lvl]);
}
SmallVector<Value> dstDcvs;
reshapeCvs(builder, loc, op.getReassociationIndices(), srcSizes,
srcDcvs, dstSizes, dstDcvs);
auto t = builder.create<InsertOp>(loc, v, reduc.front(), dstDcvs);
builder.create<sparse_tensor::YieldOp>(loc, t);
});
Value t = rewriter.create<LoadOp>(loc, foreachOp.getResult(0), true);
if (bufferTp != dstTp) {
auto dstRTT = dstTp.getRankedTensorType();
Value converted = rewriter.create<ConvertOp>(loc, dstRTT, t).getResult();
rewriter.create<DeallocTensorOp>(loc, t);
t = converted;
}
rewriter.replaceOp(op, t);
return success();
}
};
/// Sparse rewriting rule for sparse-to-dense and dense-to-sparse reshape
/// operator.
template <typename ReshapeOp>
struct ReshapeRewriter : public OpRewritePattern<ReshapeOp> {
public:
using OpRewritePattern<ReshapeOp>::OpRewritePattern;
LogicalResult matchAndRewrite(ReshapeOp op,
PatternRewriter &rewriter) const override {
Location loc = op->getLoc();
auto encDst = getSparseTensorEncoding(op.getResult().getType());
auto encSrc = getSparseTensorEncoding(op.getSrc().getType());
// Since a pure dense expansion is very cheap (change of view), for
// a sparse2dense or dense2sparse, we can simply unfuse a sparse
// conversion from the reshape operation itself.
// All other cases are handled elsewhere.
if (encDst && encSrc) {
return failure();
}
if (encSrc) {
auto rtp = getRankedTensorType(op.getSrc());
auto denseTp =
RankedTensorType::get(rtp.getShape(), rtp.getElementType());
auto convert = rewriter.create<ConvertOp>(loc, denseTp, op.getSrc());
rewriter.updateRootInPlace(op, [&]() { op->setOperand(0, convert); });
return success();
}
if (encDst) {
auto rtp = getRankedTensorType(op.getResult());
auto denseTp =
RankedTensorType::get(rtp.getShape(), rtp.getElementType());
auto reshape = rewriter.create<ReshapeOp>(loc, denseTp, op.getSrc(),
op.getReassociation());
Value convert = rewriter.create<ConvertOp>(loc, rtp, reshape);
rewriter.replaceOp(op, convert);
return success();
}
return failure();
}
};
struct ConcatenateRewriter : public OpRewritePattern<ConcatenateOp> {
using OpRewritePattern::OpRewritePattern;
LogicalResult matchAndRewrite(ConcatenateOp op,
PatternRewriter &rewriter) const override {
const Location loc = op.getLoc();
const auto dstTp = getSparseTensorType(op);
const Dimension dimRank = dstTp.getDimRank();
const Dimension conDim = op.getDimension();
SmallVector<Value> sizes;
concatSizesFromInputs(rewriter, sizes, loc, dstTp, op.getInputs(), conDim);
// %t = concatenate %s1, %s2, %s3 {dim = 1}
// ==>
// if (isSparseDst)
// if (allDense)
// %tmp = bufferization.alloc_tensor dstTp
// else
// %tmp = bufferization.alloc_tensor : unordered COO
// else
// %tmp = memref.alloc : dense tensor
// foreach in %s1 : insert d0, d1, %tmp
// foreach in %s2 : insert d0, d1 + size(s1), %tmp
// foreach in %s3 : insert d0, d1 + size(s1) + size(s2), %tmp
// %t = convert_to_dest_tensor(%tmp)
//
// NOTE: this cannot be `const` because it will be changed when
// `needTmpCOO`, but that's buried in the conditional below and
// thus not easily extracted.
auto encDst = dstTp.getEncoding();
Value dst; // Destination tensor for inserting source tensor values.
bool needTmpCOO = true;
const bool allDense = dstTp.hasEncoding() && dstTp.isAllDense();
Value annotatedDenseDst;
if (dstTp.hasEncoding()) {
bool allOrdered = false;
// When concatenating on dimension 0, and all inputs are sorted
// and have an identity dimToLvl, the concatenate will generate
// coords in lexOrder thus no need for the tmp COO buffer.
// TODO: When conDim != 0, as long as conDim is the first dimension
// in all input/output buffers, and all input/output buffers have the same
// dimToLvl, the tmp COO buffer is still unnecessary (e.g, concatenate
// CSC matrices along column).
if (!allDense && conDim == 0 && dstTp.isIdentity()) {
for (auto i : op.getInputs()) {
const auto stt = getSparseTensorType(i);
allOrdered = stt.isAllOrdered() && stt.isIdentity();
if (!allOrdered)
break;
}
}
needTmpCOO = !allDense && !allOrdered;
const RankedTensorType tp =
getBufferType(dstTp.withoutDimToLvl(), needTmpCOO);
encDst = needTmpCOO ? getSparseTensorEncoding(tp) : encDst;
SmallVector<Value> dynSizes;
getDynamicSizes(dstTp, sizes, dynSizes);
dst = rewriter.create<AllocTensorOp>(loc, tp, dynSizes).getResult();
if (allDense) {
// Create a view of the values buffer to match the unannotated dense
// tensor.
Value valuesBuffer = genToValues(rewriter, loc, dst);
Value dimCoords =
genAlloca(rewriter, loc, dimRank, rewriter.getIndexType(),
/*staticShape=*/true);
annotatedDenseDst = dst;
dst = reshapeValuesToLevels(rewriter, loc, encDst, sizes, valuesBuffer,
dimCoords);
}
} else {
// TODO: Dense buffers should be allocated/deallocated via the callback
// in BufferizationOptions.
dst = allocDenseTensor(rewriter, loc, dstTp, sizes);
}
Value offset = constantIndex(rewriter, loc, 0);
SmallVector<Value> initArgs;
if (encDst && !allDense)
initArgs.push_back(dst);
ForeachOp foreachOp;
for (Value input : op.getInputs()) {
// Build a for op for each input tensor to append new values into the
// output tensor.
foreachOp = rewriter.create<ForeachOp>(
loc, input, initArgs,
[&](OpBuilder &builder, Location loc, ValueRange dcvs, Value v,
ValueRange reduc) {
SmallVector<Value> dstLcvs(dstTp.getLvlRank());
for (Dimension d = 0; d < dimRank; d++) {
Value crd = dcvs[d];
if (d == conDim)
// Transform coordinates for the concatenating dim.
crd = builder.create<arith::AddIOp>(loc, crd, offset);
// FIXME: `toStoredDim` is deprecated
dstLcvs[toStoredDim(encDst, d)] = crd;
}
if (encDst && !allDense) {
Value cond = genIsNonzero(rewriter, loc, v);
scf::IfOp ifOp = builder.create<scf::IfOp>(
loc, TypeRange(reduc.front().getType()), cond, /*else*/ true);
builder.setInsertionPointToStart(&ifOp.getThenRegion().front());
Value t =
builder.create<InsertOp>(loc, v, reduc.front(), dstLcvs);
rewriter.create<scf::YieldOp>(loc, t);
rewriter.setInsertionPointToStart(&ifOp.getElseRegion().front());
rewriter.create<scf::YieldOp>(loc, reduc.front());
rewriter.setInsertionPointAfter(ifOp);
rewriter.create<sparse_tensor::YieldOp>(loc, ifOp.getResult(0));
} else {
builder.create<memref::StoreOp>(loc, v, dst, dstLcvs);
builder.create<sparse_tensor::YieldOp>(loc);
}
});
// Accumulates the offset. Note that only static-shaped inputs are allowed
// by concatenate op verifier, which saves us from computing the offset
// dynamically.
const auto sh = getSparseTensorType(input).getStaticDimSize(conDim);
assert(sh.has_value());
offset = rewriter.create<arith::AddIOp>(
loc, offset, constantIndex(rewriter, loc, *sh));
if (encDst && !allDense) {
dst = foreachOp.getResult(0);
initArgs[0] = dst;
}
}
// Temp variable to avoid needing to call `getRankedTensorType`
// in the three use-sites below.
const RankedTensorType dstRTT = dstTp;
if (!encDst) {
rewriter.replaceOpWithNewOp<bufferization::ToTensorOp>(op, dstRTT, dst);
} else if (allDense) {
rewriter.replaceOp(
op, rewriter.create<ConvertOp>(loc, dstRTT, annotatedDenseDst)
.getResult());
} else {
dst = rewriter.create<LoadOp>(loc, dst, true);
if (needTmpCOO) {
Value tmpCoo = dst;
dst = rewriter.create<ConvertOp>(loc, dstRTT, tmpCoo).getResult();
rewriter.create<DeallocTensorOp>(loc, tmpCoo);
}
rewriter.replaceOp(op, dst);
}
return success();
}
};
/// Sparse rewriting rule for the convert operator.
struct ConvertRewriter : public OpRewritePattern<ConvertOp> {
using OpRewritePattern::OpRewritePattern;
LogicalResult matchAndRewrite(ConvertOp op,
PatternRewriter &rewriter) const override {
auto encDst = getSparseTensorEncoding(op.getType());
auto encSrc = getSparseTensorEncoding(op.getSource().getType());
if (encDst && encSrc && !encSrc.isSlice() &&
encSrc.withoutBitWidths() == encDst.withoutBitWidths()) {
// Trivial tensor conversion and simple element type conversion is handled
// in codegen.
return failure();
}
// TODO: Add a cast before generating InsertOp.
assert(op.getSource().getType().getElementType() ==
op.getDest().getType().getElementType());
if (encSrc && encDst)
return sparse2SparseRewrite(op, rewriter);
if (encSrc && !encDst)
return sparse2DenseRewrite(op, rewriter);
if (!encSrc && encDst)
return dense2SparseRewrite(op, rewriter);
// Dense-to-dense convert is a nop and handled by canonicalization.
return failure();
}
private:
// Handles sparse constant to sparse tensor or dense tensor to sparse tensor
// conversion as follows:
// t = new sparse COO tensor
// fill t using src
// dst = convert t
//
// To fill the COO tensor from a dense tensor:
// for i1 in dim1
// ..
// for ik in dimk
// val = a[i1,..,ik]
// if val != 0
// t->add(val, [i1,..,ik], [p1,..,pk])
//
// To fill the COO tensor from a sparse constant in COO format:
// for i in range(NNZ)
// val = values[i]
// [i1,..,ik] = coordinates[i]
// t->add(val, [i1,..,ik], [p1,..,pk])
LogicalResult dense2SparseRewrite(ConvertOp op,
PatternRewriter &rewriter) const {
Location loc = op.getLoc();
Value src = op.getSource();
const auto dstTp = getSparseTensorType(op);
SmallVector<Value> sizes;
sizesFromSrc(rewriter, sizes, loc, src);
SmallVector<Value> dynSizes;
getDynamicSizes(dstTp, sizes, dynSizes);
bool fromSparseConst = false;
if (auto constOp = op.getSource().getDefiningOp<arith::ConstantOp>()) {
if (dyn_cast<SparseElementsAttr>(constOp.getValue())) {
fromSparseConst = true;
}
}
const auto encDst = dstTp.getEncoding();
// We don't need a temporary COO tensor if the destination has an identity
// ordering. Otherwise, we use the destination ordering for the temporary
// COO tensor.
// TODO: enhance foreachOp to take ordering to remove the need of a
// temporary COO tensor here.
const RankedTensorType bufferTp =
getBufferType(dstTp, !dstTp.isIdentity() && !fromSparseConst);
// Only imposes foreach order on dense constant (which will be statically
// sorted by the sparse compiler), otherwise the rotated loop sequence
// results to bad cache locality.
const AffineMapAttr foreachOrder =
(!dstTp.isIdentity() && fromSparseConst)
? AffineMapAttr::get(dstTp.getExpandedDimToLvl())
: nullptr;
// TODO: This assertion is to match the behavior from before we merged
// dimOrdering and higherOrdering into dimToLvl. Although the above
// can construct `foreachOrder` for non-permutations, it's not clear
// that the `foreachOp` below actually supports non-permutations.
assert(!foreachOrder || dstTp.isPermutation());
auto buffer =
rewriter.create<AllocTensorOp>(loc, bufferTp, dynSizes).getResult();
auto foreachOp = rewriter.create<ForeachOp>(
loc, src, buffer, foreachOrder,
[&](OpBuilder &builder, Location loc, ValueRange dcvs, Value v,
ValueRange reduc) {
Value input = reduc.front();
const Dimension dimRank = dstTp.getDimRank();
const Level lvlRank = dstTp.getLvlRank();
SmallVector<Value> lcvs(lvlRank);
for (Dimension d = 0; d < dimRank; d++)
// FIXME: `toStoredDim` is deprecated
lcvs[toStoredDim(encDst, d)] = dcvs[d];
if (fromSparseConst) {
input = builder.create<InsertOp>(loc, v, input, lcvs);
} else {
Value cond = genIsNonzero(builder, loc, v);
auto ifOp = builder.create<scf::IfOp>(
loc, TypeRange(input.getType()), cond, /*else*/ true);
builder.setInsertionPointToStart(&ifOp.getThenRegion().front());
Value insert = builder.create<InsertOp>(loc, v, input, lcvs);
builder.create<scf::YieldOp>(loc, insert);
builder.setInsertionPointToStart(&ifOp.getElseRegion().front());
builder.create<scf::YieldOp>(loc, input);
builder.setInsertionPointAfter(ifOp);
input = ifOp.getResult(0);
}
builder.create<sparse_tensor::YieldOp>(loc, input);
});
rewriter.setInsertionPointAfter(op);
src = rewriter.create<LoadOp>(loc, foreachOp.getResult(0), true);
if (bufferTp != dstTp) {
rewriter.replaceOpWithNewOp<ConvertOp>(op, dstTp.getRankedTensorType(),
src);
rewriter.create<DeallocTensorOp>(loc, src);
} else {
rewriter.replaceOp(op, src);
}
return success();
}
// Handles sparse tensor to dense tensor conversion as follows:
// dst = new dense tensor;
// foreach elemment in src
// dst[element.coords] = element.value
LogicalResult sparse2DenseRewrite(ConvertOp op,
PatternRewriter &rewriter) const {
Location loc = op->getLoc();
RankedTensorType dstTp = getRankedTensorType(op);
Value src = op.getSource();
RankedTensorType srcTp = getRankedTensorType(src);
SmallVector<Value> sizes;
sizesForTensor(rewriter, sizes, loc, srcTp, src);
Value dst = allocDenseTensor(rewriter, loc, dstTp, sizes);
Block *insertionBlock = rewriter.getInsertionBlock();
bool noEscape = bufferization::allocationDoesNotEscape(op->getOpResult(0));
rewriter.create<ForeachOp>(loc, src, std::nullopt,
[&](OpBuilder &builder, Location loc,
ValueRange args, Value v, ValueRange reduc) {
builder.create<memref::StoreOp>(loc, v, dst,
args);
builder.create<sparse_tensor::YieldOp>(loc);
});
rewriter.replaceOpWithNewOp<bufferization::ToTensorOp>(op, dstTp, dst);
// Deallocate the buffer.
if (noEscape) {
rewriter.setInsertionPoint(insertionBlock->getTerminator());
deallocDenseTensor(rewriter, loc, dst);
}
return success();
}
// Handles sparse tensor to sparse tensor conversion as follows:
// if src is not COO
// construct a COO to represent the src
// sort the src COO
// foreach elemment in the sorted src COO
// insert element to dst
LogicalResult sparse2SparseRewrite(ConvertOp op,
PatternRewriter &rewriter) const {
const Location loc = op->getLoc();
// These two variables cannot be `const` because they're conditionally
// changed below. Ideally we'd use `SparseTensorType` for `srcRTT`;
// however that class's copy-ctor is implicitly deleted.
Value src = op.getSource();
auto srcRTT = getRankedTensorType(src);
const auto dstTp = getSparseTensorType(op);
const auto encDst = dstTp.getEncoding();
const Level dstLvlRank = dstTp.getLvlRank();
const Dimension dimRank = dstTp.getDimRank();
// This assertion should be guaranteed by validity of the op,
// but just for paranoia's sake.
assert(static_cast<Dimension>(srcRTT.getRank()) == dimRank);
SmallVector<Value> srcSizes;
sizesForTensor(rewriter, srcSizes, loc, srcRTT, src);
Value tmpCoo = Value();
Value nnz = rewriter.create<NumberOfEntriesOp>(loc, src);
// We need a tmp COO buffer if and only if
// 1. the src tensor is not a COO and
// 2. the src tensor is not ordered in the same way as the target
// tensor (e.g., src tensor is not ordered or src tensor haves a different
// dimToLvl).
if (const SparseTensorType srcTp(srcRTT);
!(srcTp.isAllOrdered() && srcTp.hasSameDimToLvl(dstTp))) {
// Construct a COO tensor from the src tensor.
// TODO: there may be cases for which more efficiently without
// going through an intermediate COO, such as cases that only change
// the overhead types.
SmallVector<Value> dynSrcSizes;
getDynamicSizes(srcRTT, srcSizes, dynSrcSizes);
srcRTT = getCOOType(srcTp.withDimToLvl(dstTp), /*ordered=*/false);
// Ensure that mutating `srcRTT` didn't invalidate `dimRank`.
assert(static_cast<Dimension>(srcRTT.getRank()) == dimRank);
tmpCoo = rewriter
.create<AllocTensorOp>(loc, srcRTT, dynSrcSizes, Value(),
/*sizeHint=*/nnz, Attribute())
.getResult();
auto foreachOp = rewriter.create<ForeachOp>(
loc, src, tmpCoo,
[&](OpBuilder &builder, Location loc, ValueRange dcvs, Value v,
ValueRange reduc) {
SmallVector<Value> dstLcvs(dstLvlRank);
for (Dimension d = 0; d < dimRank; d++) {
// FIXME: `toStoredDim` is deprecated
Level l = toStoredDim(encDst, d);
dstLcvs[l] = dcvs[d];
}
auto t = builder.create<InsertOp>(loc, v, reduc.front(), dstLcvs);
builder.create<sparse_tensor::YieldOp>(loc, t);
});
src = rewriter.create<LoadOp>(loc, foreachOp.getResult(0), true);
}
// Now that the conditional is done, we can use `SparseTensorType`.
const SparseTensorType srcTp(srcRTT);
// Only need to sort if the srcTp is not already sorted (we faithfully take
// the guarantee from the sparse tensor encoding).
if (!srcTp.isAllOrdered()) {
// Retrieve the values-array.
Value y = genToValues(rewriter, loc, src);
const auto encSrc = srcTp.getEncoding();
// Sort the COO tensor so that its elements are ordered via increasing
// coordinates for the storage ordering of the dst tensor. Use SortCoo
// if the COO tensor has the same ordering as the dst tensor.
if (dimRank > 1 && srcTp.hasSameDimToLvl(dstTp)) {
Value xs = genToCoordinatesBuffer(rewriter, loc, src);
rewriter.create<SortCooOp>(
loc, nnz, xs, ValueRange{y}, rewriter.getIndexAttr(dimRank),
rewriter.getIndexAttr(0), SparseTensorSortKind::HybridQuickSort);
} else {
// Gather the coordinates-arrays in the dst tensor storage order.
SmallVector<Value> xs(dstLvlRank);
const Level srcLvlRank = srcTp.getLvlRank();
for (Level srcLvl = 0; srcLvl < srcLvlRank; srcLvl++) {
// FIXME: `toOrigDim` is deprecated
Dimension dim = toOrigDim(encSrc, srcLvl);
// FIXME: `toStoredDim` is deprecated
Level dstLvl = toStoredDim(encDst, dim);
xs[dstLvl] =
genToCoordinates(rewriter, loc, src, srcLvl, /*cooStart=*/0);
}
rewriter.create<SortOp>(loc, nnz, xs, ValueRange{y},
SparseTensorSortKind::HybridQuickSort);
}
}
// For each element in the COO tensor, insert the element to the dst tensor.
SmallVector<Value> dynDstSizes;
getDynamicSizes(dstTp, srcSizes, dynDstSizes);
Value dst = rewriter
.create<AllocTensorOp>(loc, dstTp.getRankedTensorType(),
dynDstSizes, Value(),
/*sizeHint=*/nnz, Attribute())
.getResult();
SmallVector<Value> dstLcvs(dstLvlRank);
auto foreachOp = rewriter.create<ForeachOp>(
loc, src, dst,
[&](OpBuilder &builder, Location loc, ValueRange dcvs, Value v,
ValueRange reduc) {
for (Dimension d = 0; d < dimRank; d++) {
// FIXME: `toStoredDim` is deprecated
Level l = toStoredDim(encDst, d);
dstLcvs[l] = dcvs[d];
}
auto t = builder.create<InsertOp>(loc, v, reduc.front(), dstLcvs);
builder.create<sparse_tensor::YieldOp>(loc, t);
});
// Release the temporary COO if it is created. Note that tmpCoo is
// invalidated due to foreach and updated to src.
if (tmpCoo)
rewriter.create<DeallocTensorOp>(loc, src);
// Directly replace op with dst results in bufferization error message
// "sparse tensor allocation should not escape function".
// As such, we insert a trivial tensor convert which will be removed by
// codegen.
rewriter.setInsertionPointAfter(op);
auto t = rewriter.create<LoadOp>(loc, foreachOp.getResult(0), true);
rewriter.replaceOpWithNewOp<ConvertOp>(op, dstTp.getRankedTensorType(), t);
return success();
}
};
/// Sparse rewriting rule for the foreach operator.
struct ForeachRewriter : public OpRewritePattern<ForeachOp> {
public:
using OpRewritePattern::OpRewritePattern;
LogicalResult matchAndRewrite(ForeachOp op,
PatternRewriter &rewriter) const override {
auto loc = op.getLoc();
Value input = op.getTensor();
SmallVector<Value> reduc = op.getInitArgs();
const auto stt = getSparseTensorType(input);
const Dimension dimRank = stt.getDimRank();
const Level lvlRank = stt.getLvlRank();
// Special-case: for each over a sparse constant uses its own rewriting
// rule.
if (auto constOp = input.getDefiningOp<arith::ConstantOp>()) {
if (auto attr = dyn_cast<SparseElementsAttr>(constOp.getValue())) {
return genForeachOnSparseConstant(op, rewriter, attr);
}
}
// Otherwise, use loop emitter to generate loops.
const auto enc = stt.getEncoding();
// 1. Generates loop for the sparse input.
LoopEmitter loopEmitter(
ValueRange{input},
StringAttr::get(getContext(), ForeachOp::getOperationName()));
loopEmitter.initializeLoopEmit(rewriter, loc);
for (Level l = 0; l < lvlRank; l++) {
// TODO: provide utility function for loop sequences that only contains
// one for loop?
const SmallVector<TensorLevel, 1> tidLvls{
loopEmitter.makeTensorLevel(0, l)};
loopEmitter.enterNewLoopSeq(rewriter, loc, tidLvls);
// Note that reduc will be taken care of by loop emitter and get updated
// in place.
loopEmitter.enterLoopOverTensorAtLvl(rewriter, loc, tidLvls, reduc);
}
SmallVector<Value> lcvs;
lcvs.reserve(lvlRank);
loopEmitter.getLoopIVs(lcvs);
if (op.getOrder()) {
// FIXME: There is some dim/lvl confusion here since `dimRank != lvlRank`
SmallVector<Value> dcvs = lcvs; // keep a copy
for (Dimension d = 0; d < dimRank; d++) {
auto l = op.getOrder()->getDimPosition(d);
lcvs[l] = dcvs[d];
}
}
Value vals = loopEmitter.getValBuffer()[0];
Value pos = loopEmitter.getPosits()[0].back();
// Loads the value from sparse tensor using position-index;
// loads the value from dense tensor using coords.
Value val = enc ? rewriter.create<memref::LoadOp>(loc, vals, pos)
: rewriter.create<memref::LoadOp>(loc, vals, lcvs);
// 2. Inline the block in the foreach operator.
Block *srcBlock = op.getBody();
// Remap coordinates.
SmallVector<Value> args;
for (Dimension d = 0; d < dimRank; d++) {
// FIXME: `toStoredDim` is deprecated
Value dimCrd = lcvs[toStoredDim(enc, d)];
args.push_back(dimCrd);
}
// Remap value.
args.push_back(val);
// Remap reduction variables.
args.append(reduc);
// Remove sparse_tensor.yield.
SmallVector<Value> reducValue = srcBlock->getTerminator()->getOperands();
rewriter.eraseOp(srcBlock->getTerminator());
// Inline body.
if (!reducValue.empty()) {
rewriter.mergeBlocks(srcBlock, rewriter.getBlock(), args);
} else {
// This is annoying, since scf.for inserts a implicit yield op when
// there is no reduction variable upon creation, in this case we need to
// merge the block *before* the yield op.
rewriter.inlineBlockBefore(srcBlock, &*rewriter.getInsertionPoint(),
args);
}
for (Dimension d = 0; d < dimRank; d++) {
// Link the reduction chain. Note that loop emitter update the reducValue
// in place.
loopEmitter.exitCurrentLoop(rewriter, loc, reducValue);
loopEmitter.exitCurrentLoopSeq(rewriter, loc);
}
// Replace the foreach operator with the value returned by the outtermost
// for loop.
rewriter.replaceOp(op, reducValue);
return success();
}
};
/// Sparse rewriting rule for the new operator.
struct NewRewriter : public OpRewritePattern<NewOp> {
using OpRewritePattern::OpRewritePattern;
LogicalResult matchAndRewrite(NewOp op,
PatternRewriter &rewriter) const override {
Location loc = op.getLoc();
const auto dstTp = getSparseTensorType(op.getResult());
const auto encDst = dstTp.getEncoding();
if (!dstTp.hasEncoding() || getCOOStart(encDst) == 0)
return failure();
// Implement the NewOp as follows:
// %orderedCoo = sparse_tensor.new %filename
// %t = sparse_tensor.convert %orderedCoo
RankedTensorType cooTp = getCOOType(dstTp, /*ordered=*/true);
Value cooTensor = rewriter.create<NewOp>(loc, cooTp, op.getSource());
Value convert = rewriter.replaceOpWithNewOp<ConvertOp>(
op, dstTp.getRankedTensorType(), cooTensor);
// Release the ordered COO tensor.
rewriter.setInsertionPointAfterValue(convert);
rewriter.create<DeallocTensorOp>(loc, cooTensor);
return success();
}
};
struct OutRewriter : public OpRewritePattern<OutOp> {
using OpRewritePattern::OpRewritePattern;
LogicalResult matchAndRewrite(OutOp op,
PatternRewriter &rewriter) const override {
Location loc = op.getLoc();
// Calculate NNZ.
Value src = op.getTensor();
Value nnz = rewriter.create<NumberOfEntriesOp>(loc, src);
// Allocate a temporary buffer for storing dimension-sizes/coordinates.
const auto srcTp = getSparseTensorType(src);
const Dimension dimRank = srcTp.getDimRank();
Type indexTp = rewriter.getIndexType();
Value dimSizes = genAlloca(rewriter, loc, dimRank, indexTp);
// Generate code to calculate dimension size values and store the values to
// the buffer.
SmallVector<Value> dims;
sizesForTensor(rewriter, dims, loc, srcTp, src);
for (Dimension d = 0; d < dimRank; d++) {
rewriter.create<memref::StoreOp>(loc, dims[d], dimSizes,
constantIndex(rewriter, loc, d));
}
// Create a sparse tensor writer and output meta data.
Type opaqueTp = getOpaquePointerType(rewriter);
Value writer =
createFuncCall(rewriter, loc, "createSparseTensorWriter", {opaqueTp},
{op.getDest()}, EmitCInterface::Off)
.getResult(0);
Value rankValue = constantIndex(rewriter, loc, dimRank);
createFuncCall(rewriter, loc, "outSparseTensorWriterMetaData", {},
{writer, rankValue, nnz, dimSizes}, EmitCInterface::On);
Value dimCoords = dimSizes; // Reuse the dimSizes buffer for dimCoords.
Type eltTp = srcTp.getElementType();
SmallString<29> outNextFuncName{"outSparseTensorWriterNext",
primaryTypeFunctionSuffix(eltTp)};
Value value = genAllocaScalar(rewriter, loc, eltTp);
ModuleOp module = op->getParentOfType<ModuleOp>();
// For each element in the source tensor, output the element.
rewriter.create<ForeachOp>(
loc, src, std::nullopt,
[&](OpBuilder &builder, Location loc, ValueRange dcvs, Value v,
ValueRange reduc) {
for (Dimension d = 0; d < dimRank; d++) {
rewriter.create<memref::StoreOp>(loc, dcvs[d], dimCoords,
constantIndex(builder, loc, d));
}
rewriter.create<memref::StoreOp>(loc, v, value);
SmallVector<Value> operands{writer, rankValue, dimCoords, value};
FlatSymbolRefAttr fn = getFunc(module, outNextFuncName, {}, operands,
EmitCInterface::On);
builder.create<func::CallOp>(loc, TypeRange(), fn, operands);
builder.create<sparse_tensor::YieldOp>(loc);
});
// Release the writer.
createFuncCall(rewriter, loc, "delSparseTensorWriter", {}, {writer},
EmitCInterface::Off);
rewriter.eraseOp(op);
return success();
}
};
} // namespace
//===---------------------------------------------------------------------===//
// Methods that add patterns described in this file to a pattern list.
//===---------------------------------------------------------------------===//
void mlir::populatePreSparsificationRewriting(RewritePatternSet &patterns) {
patterns.add<FoldInvariantYield, FuseSparseMultiplyOverAdd, FuseTensorCast,
GenSemiRingReduction>(patterns.getContext());
}
void mlir::populatePostSparsificationRewriting(RewritePatternSet &patterns,
bool enableRT,
bool enableForeach,
bool enableConvert) {
patterns.add<ReshapeRewriter<tensor::ExpandShapeOp>,
ReshapeRewriter<tensor::CollapseShapeOp>, TensorReshapeRewriter>(
patterns.getContext());
if (enableForeach)
patterns.add<ForeachRewriter>(patterns.getContext());
// TODO: If RT not enabled, rewrite concatenate ops, etc here.
if (!enableRT) {
patterns.add<ConcatenateRewriter, NewRewriter, OutRewriter,
Sparse2SparseReshapeRewriter<tensor::ExpandShapeOp>,
Sparse2SparseReshapeRewriter<tensor::CollapseShapeOp>>(
patterns.getContext());
if (enableConvert)
patterns.add<ConvertRewriter>(patterns.getContext());
}
}
Reference in New Issue View Git Blame Copy Permalink