#include "mlir/Dialect/Linalg/TransformOps/LinalgTransformOps.h"
#include "mlir/AsmParser/AsmParser.h"
#include "mlir/Dialect/Affine/IR/AffineOps.h"
#include "mlir/Dialect/Arith/IR/Arith.h"
#include "mlir/Dialect/GPU/IR/GPUDialect.h"
#include "mlir/Dialect/Linalg/IR/Linalg.h"
#include "mlir/Dialect/Linalg/TransformOps/Syntax.h"
#include "mlir/Dialect/Linalg/Transforms/Hoisting.h"
#include "mlir/Dialect/Linalg/Transforms/Transforms.h"
#include "mlir/Dialect/Linalg/Utils/Utils.h"
#include "mlir/Dialect/SCF/Transforms/TileUsingInterface.h"
#include "mlir/Dialect/Tensor/IR/Tensor.h"
#include "mlir/Dialect/Tensor/Utils/Utils.h"
#include "mlir/Dialect/Transform/IR/TransformDialect.h"
#include "mlir/Dialect/Transform/IR/TransformInterfaces.h"
#include "mlir/Dialect/Transform/IR/TransformOps.h"
#include "mlir/Dialect/Transform/IR/TransformTypes.h"
#include "mlir/Dialect/Transform/Utils/Utils.h"
#include "mlir/Dialect/Utils/IndexingUtils.h"
#include "mlir/Dialect/Utils/StaticValueUtils.h"
#include "mlir/Dialect/Vector/Transforms/LoweringPatterns.h"
#include "mlir/Dialect/Vector/Transforms/VectorRewritePatterns.h"
#include "mlir/IR/PatternMatch.h"
#include "mlir/IR/TypeUtilities.h"
#include "mlir/Interfaces/TilingInterface.h"
#include "mlir/Support/LLVM.h"
#include "mlir/Support/TypeID.h"
#include "mlir/Transforms/GreedyPatternRewriteDriver.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/ADT/ScopeExit.h"
#include "llvm/ADT/TypeSwitch.h"
#include "llvm/Support/Debug.h"
#include <type_traits>
#include "mlir/Dialect/Linalg/TransformOps/LinalgTransformOpsEnums.cpp.inc"
#include "mlir/Dialect/Linalg/TransformOps/LinalgTransformOps.cpp.inc"

Macros
#define	DEBUG_TYPE "linalg-transforms"

#define	DBGS() (llvm::dbgs() << "[" DEBUG_TYPE "]: ")

#define	DBGSNL() (llvm::dbgs() << "\n")

#define	DOWNSCALE(trans)

#define	DOWNSCALE_CALL(a, b) DownscaleSizeOneWindowed2DConvolution<a, b>

#define	DOWNSCALE_NORMAL(a, b) DOWNSCALE(DOWNSCALE_CALL(a, b))

#define	GET_OP_CLASSES

Functions
template<typename PatternTy , typename... Args>
static FailureOr< LinalgOp >	tryApply (Operation *operation, Args &&...args)
	Attempts to apply the pattern specified as template argument to the given operation. More...

static DiagnosedSilenceableFailure	unpackSingleIndexResultPayloadOperations (transform::TransformState &state, TransformOpInterface transformOp, SmallVector< OpFoldResult > &result, ArrayRef< OpFoldResult > ofrs)
	Assuming that `ofr` is an index attr or a transform dialect handle mapped to exactly one op with one index result, return that value. More...

static DiagnosedSilenceableFailure	unpackSingleIndexResultPayloadOperations (transform::TransformState &state, TransformOpInterface transformOp, SmallVector< OpFoldResult > &result, Value packedHandle)

template<typename Range >
static LogicalResult	applyTilingToAll (RewriterBase &rewriter, Operation *transformOp, Range &&payloadOps, unsigned numLoops, transform::TransformResults &transformResults, function_ref< FailureOr< scf::SCFTileAndFuseResult >(TilingInterface)> applyFn)
	Apply a tiling transformation to all payload ops and store both the tiled operation as well as the created tile loops. More...

static Operation *	replaceForAllWithNewSignature (RewriterBase &rewriter, Diagnostic &diag, Operation producerOp, Operation containingOp, TilingResult &tileAndFuseResult, int64_t resultNumber, SmallVector< OpFoldResult > &offsets, SmallVector< OpFoldResult > &sizes)
	Add new operands to the forall op for users of the producerOp that are dominated by the containing scf.forall op. More...

static std::tuple< SmallVector< Operation * >, Operation * >	tileAndFuseFirstExtractUse (RewriterBase &rewriter, Diagnostic &diag, Operation producerOp, Operation containingOp)
	Find the first "extract" user of `producerOp` and tile it right before its use. More...

static SmallVector< Operation * >	tileAndFuseFirstExtractUseThroughContainingOpBlockArgument (RewriterBase &rewriter, Diagnostic &diag, Operation producerOp, Operation containingOp)
	First, find the first "scf::ForallOp" user of `producerOp` and ensure it is exactly the `containingOp`, otherwise bail. More...

static Operation *	cloneAndFuseFirstUse (RewriterBase &rewriter, Diagnostic &diag, Operation producerOp, Operation containingOp)

static void	forciblyReplaceReferencedPayloadOperation (TransformState &state, Operation payload, Operation replacement)
	Replaces `payload` with `replacement` in all handles stored in the state. More...

static void	printMultitileSizesTypes (OpAsmPrinter &printer, Operation *op, Type targetType, Type lowSizeType, Type, Type)

static ParseResult	parseMultitileSizesTypes (OpAsmParser &parser, Type &targetType, Type &lowSizeType, Type &highSizeType, Type &splitPointType)

static FailureOr< PackResult >	packMatmulGreedily (RewriterBase &rewriter, LinalgOp linalgOp, ArrayRef< OpFoldResult > mnkPackedSizes, ArrayRef< int64_t > mnkPaddedSizesNextMultipleOf, ArrayRef< int64_t > mnkOrder)
	Pack a LinalgOp by greedily inferring matmul dimensions (m, n, k) where m and n are proper parallel dimensions and k is a proper reduction dimension. More...

template<typename RelayoutOpTy >
bool	isValidPackingPermutation (RelayoutOpTy op, ArrayRef< int64_t > permutation, OuterOrInnerPerm outerOrInnerPerm=OuterOrInnerPerm::Outer)
	Return true if `permutation` is a valid permutation of the `outer_dims_perm` (case OuterOrInnerPerm::Outer) or `inner_dims_pos` (OuterOrInnerPerm::Inner) of the `tensor.pack` or `tensor.unpack` `op.` More...

ParseResult	parseOptionalInterchange (OpAsmParser &parser, OperationState &result)

void	printOptionalInterchange (OpAsmPrinter &p, ArrayRef< int64_t > interchangeVals)

template<typename OpTy >
DiagnosedSilenceableFailure	doit (RewriterBase &rewriter, OpTy target, transform::ApplyToEachResultList &results, transform::TransformState &state)

Macro Definition Documentation

◆ DBGS

#define DBGS ( ) (llvm::dbgs() << "[" DEBUG_TYPE "]: ")

Definition at line 50 of file LinalgTransformOps.cpp.

◆ DBGSNL

#define DBGSNL ( ) (llvm::dbgs() << "\n")

Definition at line 51 of file LinalgTransformOps.cpp.

◆ DEBUG_TYPE

#define DEBUG_TYPE "linalg-transforms"

Definition at line 49 of file LinalgTransformOps.cpp.

◆ DOWNSCALE

#define DOWNSCALE ( trans )

Value:

  {                                                                            \
    FailureOr<LinalgOp> res = tryApply<trans>(target);                         \
    if (succeeded(res)) {                                                      \
      results.push_back(*res);                                                 \
      return DiagnosedSilenceableFailure::success();                           \
    }                                                                          \
  }

◆ DOWNSCALE_CALL

#define DOWNSCALE_CALL	(	a,
		b
	)	DownscaleSizeOneWindowed2DConvolution<a, b>

◆ DOWNSCALE_NORMAL

#define DOWNSCALE_NORMAL	(	a,
		b
	)	DOWNSCALE(DOWNSCALE_CALL(a, b))

◆ GET_OP_CLASSES

#define GET_OP_CLASSES

Definition at line 3278 of file LinalgTransformOps.cpp.

Function Documentation

◆ applyTilingToAll()

template<typename Range >

static LogicalResult applyTilingToAll	(	RewriterBase &	rewriter,
		Operation *	transformOp,
		Range &&	payloadOps,
		unsigned	numLoops,
		transform::TransformResults &	transformResults,
		function_ref< FailureOr< scf::SCFTileAndFuseResult >(TilingInterface)>	applyFn
	)

static

Apply a tiling transformation to all payload ops and store both the tiled operation as well as the created tile loops.

Definition at line 212 of file LinalgTransformOps.cpp.

References mlir::Operation::emitError(), mlir::failed(), mlir::failure(), mlir::Operation::getOpResult(), mlir::RewriterBase::replaceOp(), mlir::transform::TransformResults::set(), mlir::OpBuilder::setInsertionPoint(), and mlir::success().

◆ cloneAndFuseFirstUse()

static Operation* cloneAndFuseFirstUse	(	RewriterBase &	rewriter,
		Diagnostic &	diag,
		Operation *	producerOp,
		Operation *	containingOp
	)

static

Definition at line 615 of file LinalgTransformOps.cpp.

References mlir::OpBuilder::clone(), DBGS, diag(), mlir::IROperand< DerivedT, IRValueT >::get(), mlir::Operation::getLoc(), mlir::Operation::getOpResults(), mlir::detail::IROperandBase::getOwner(), mlir::Operation::isProperAncestor(), mlir::OpBuilder::setInsertionPoint(), and mlir::RewriterBase::updateRootInPlace().

◆ doit()

template<typename OpTy >

DiagnosedSilenceableFailure doit	(	RewriterBase &	rewriter,
		OpTy	target,
		transform::ApplyToEachResultList &	results,
		transform::TransformState &	state
	)

Definition at line 3220 of file LinalgTransformOps.cpp.

References mlir::OpBuilder::create(), mlir::transform::ApplyToEachResultList::push_back(), mlir::RewriterBase::replaceOpWithNewOp(), mlir::OpBuilder::setInsertionPoint(), and mlir::DiagnosedSilenceableFailure::success().

Referenced by llvm::cast_convert_val< T, ::mlir::Dialect *, ::mlir::Dialect * >::doit().

◆ forciblyReplaceReferencedPayloadOperation()

static void forciblyReplaceReferencedPayloadOperation	(	TransformState &	state,
		Operation *	payload,
		Operation *	replacement
	)

static

Replaces payload with replacement in all handles stored in the state.

MUST NOT be used except for the case immediately below.

Definition at line 689 of file LinalgTransformOps.cpp.

◆ isValidPackingPermutation()

template<typename RelayoutOpTy >

bool isValidPackingPermutation	(	RelayoutOpTy	op,
		ArrayRef< int64_t >	permutation,
		OuterOrInnerPerm	outerOrInnerPerm = `OuterOrInnerPerm::Outer`
	)

Return true if permutation is a valid permutation of the outer_dims_perm (case OuterOrInnerPerm::Outer) or inner_dims_pos (OuterOrInnerPerm::Inner) of the tensor.pack or tensor.unpack op.

This is the case when thepermutationrank matches the rank expected by opandpermutationis itself a permutation vector. Return true if eitheroporpermutation` are empty to allow a simpler polymorphic implementation.

Definition at line 1420 of file LinalgTransformOps.cpp.

◆ packMatmulGreedily()

static FailureOr<PackResult> packMatmulGreedily	(	RewriterBase &	rewriter,
		LinalgOp	linalgOp,
		ArrayRef< OpFoldResult >	mnkPackedSizes,
		ArrayRef< int64_t >	mnkPaddedSizesNextMultipleOf,
		ArrayRef< int64_t >	mnkOrder
	)

static

Pack a LinalgOp by greedily inferring matmul dimensions (m, n, k) where m and n are proper parallel dimensions and k is a proper reduction dimension.

Packing occurs by rewriting the op as a linalg.generic and calling linalg::pack by mnkPackedSizes. The order of the packed dimensions is customizable: the mnkOrder is a permutation of {0, 1, 2} to reorder {m, n, k} into one of the 8 possible forms. The outer dimensions of the operands are not permuted at this time, this is left for future work.

Definition at line 1217 of file LinalgTransformOps.cpp.

References mlir::bindDims(), mlir::bindSymbols(), mlir::AffineExpr::ceilDiv(), mlir::computePermutationVector(), DBGS, DBGSNL, mlir::failed(), mlir::linalg::generalizeNamedOp(), mlir::Builder::getContext(), mlir::Builder::getIndexAttr(), mlir::linalg::inferMatmulDims(), mlir::linalg::interchangeGenericOp(), mlir::isPermutationVector(), mlir::affine::makeComposedFoldedAffineApply(), mlir::RewriterBase::notifyMatchFailure(), mlir::linalg::pack(), mlir::Range::size, and mlir::succeeded().

◆ parseMultitileSizesTypes()

static ParseResult parseMultitileSizesTypes	(	OpAsmParser &	parser,
		Type &	targetType,
		Type &	lowSizeType,
		Type &	highSizeType,
		Type &	splitPointType
	)

static

Definition at line 1011 of file LinalgTransformOps.cpp.

References mlir::AsmParser::emitError(), mlir::failed(), mlir::failure(), mlir::AsmParser::getCurrentLocation(), mlir::AsmParser::parseType(), and mlir::success().

◆ parseOptionalInterchange()

ParseResult parseOptionalInterchange	(	OpAsmParser &	parser,
		OperationState &	result
	)

Definition at line 2490 of file LinalgTransformOps.cpp.

References mlir::OperationState::addAttribute(), mlir::AsmParser::emitError(), mlir::failed(), mlir::AsmParser::getNameLoc(), mlir::detail::DenseArrayAttrImpl< T >::parse(), mlir::AsmParser::parseEqual(), mlir::AsmParser::parseKeyword(), mlir::AsmParser::parseOptionalLBrace(), mlir::AsmParser::parseRBrace(), mlir::succeeded(), and mlir::success().

◆ printMultitileSizesTypes()

static void printMultitileSizesTypes	(	OpAsmPrinter &	printer,
		Operation *	op,
		Type	targetType,
		Type	lowSizeType,
		Type	,
		Type
	)

static

Definition at line 1005 of file LinalgTransformOps.cpp.

◆ printOptionalInterchange()

void printOptionalInterchange	(	OpAsmPrinter &	p,
		ArrayRef< int64_t >	interchangeVals
	)

Definition at line 2505 of file LinalgTransformOps.cpp.

◆ replaceForAllWithNewSignature()

static Operation* replaceForAllWithNewSignature	(	RewriterBase &	rewriter,
		Diagnostic &	diag,
		Operation *	producerOp,
		Operation *	containingOp,
		TilingResult &	tileAndFuseResult,
		int64_t	resultNumber,
		SmallVector< OpFoldResult > &	offsets,
		SmallVector< OpFoldResult > &	sizes
	)

static

Add new operands to the forall op for users of the producerOp that are dominated by the containing scf.forall op.

Definition at line 354 of file LinalgTransformOps.cpp.

References mlir::OpBuilder::create(), mlir::DominanceInfo::dominates(), mlir::RewriterBase::eraseBlock(), mlir::Operation::getResult(), mlir::Value::getUsers(), mlir::Operation::isAncestor(), mlir::RewriterBase::replaceUsesWithIf(), and mlir::OpBuilder::setInsertionPoint().

Referenced by tileAndFuseFirstExtractUse().

◆ tileAndFuseFirstExtractUse()

static std::tuple<SmallVector<Operation >, Operation > tileAndFuseFirstExtractUse	(	RewriterBase &	rewriter,
		Diagnostic &	diag,
		Operation *	producerOp,
		Operation *	containingOp
	)

static

Find the first "extract" user of producerOp and tile it right before its use.

The tiled op is fused under the containingOp. Return this fused op on success or nullptr if anything fails. If tiled op has uses that are dominated by containingOp, return a new containingOp with results of the fused op appended to results of the containingOp or nullptr if there are no dominated uses.

Definition at line 437 of file LinalgTransformOps.cpp.

References DBGS, diag(), mlir::failed(), mlir::Operation::getLoc(), replaceForAllWithNewSignature(), mlir::RewriterBase::replaceOp(), mlir::OpBuilder::setInsertionPoint(), and mlir::succeeded().

◆ tileAndFuseFirstExtractUseThroughContainingOpBlockArgument()

static SmallVector<Operation *> tileAndFuseFirstExtractUseThroughContainingOpBlockArgument	(	RewriterBase &	rewriter,
		Diagnostic &	diag,
		Operation *	producerOp,
		Operation *	containingOp
	)

static

First, find the first "scf::ForallOp" user of producerOp and ensure it is exactly the containingOp, otherwise bail.

Then, find the first "extract" user of the tied block argument and tile it right before its "extract" use. The tiled op is fused under the containingOp. Return this fused op on success or nullptr if anything fails.

Definition at line 513 of file LinalgTransformOps.cpp.

References mlir::OpBuilder::clone(), DBGS, diag(), mlir::RewriterBase::eraseOp(), mlir::failed(), mlir::IROperand< DerivedT, IRValueT >::get(), mlir::Operation::getLoc(), mlir::OpOperand::getOperandNumber(), mlir::tensor::getOrCreateDestinations(), mlir::Value::getUsers(), mlir::IRMapping::map(), mlir::RewriterBase::replaceOp(), mlir::OpBuilder::setInsertionPoint(), mlir::Operation::setOperand(), mlir::succeeded(), and mlir::RewriterBase::updateRootInPlace().

◆ tryApply()

template<typename PatternTy , typename... Args>

static FailureOr<LinalgOp> tryApply	(	Operation *	operation,
		Args &&...	args
	)

static

Attempts to apply the pattern specified as template argument to the given operation.

The pattern is expected to have a returningMatchAndRewrite function that returns the "main" result or failure. Returns failure if the pattern failed to apply. Extra arguments are forwarded to the pattern constructor.

Definition at line 59 of file LinalgTransformOps.cpp.

◆ unpackSingleIndexResultPayloadOperations() [1/2]

static DiagnosedSilenceableFailure unpackSingleIndexResultPayloadOperations	(	transform::TransformState &	state,
		TransformOpInterface	transformOp,
		SmallVector< OpFoldResult > &	result,
		ArrayRef< OpFoldResult >	ofrs
	)

static

Assuming that ofr is an index attr or a transform dialect handle mapped to exactly one op with one index result, return that value.

Definition at line 86 of file LinalgTransformOps.cpp.

References diag(), mlir::Value::getLoc(), and mlir::transform::TransformState::getPayloadOps().

◆ unpackSingleIndexResultPayloadOperations() [2/2]

static DiagnosedSilenceableFailure unpackSingleIndexResultPayloadOperations	(	transform::TransformState &	state,
		TransformOpInterface	transformOp,
		SmallVector< OpFoldResult > &	result,
		Value	packedHandle
	)

static

Definition at line 126 of file LinalgTransformOps.cpp.

Macros

Functions

Macro Definition Documentation

◆ DBGS

◆ DBGSNL

◆ DEBUG_TYPE

◆ DOWNSCALE

◆ DOWNSCALE_CALL

◆ DOWNSCALE_NORMAL

◆ GET_OP_CLASSES

Function Documentation

◆ applyTilingToAll()

◆ cloneAndFuseFirstUse()

◆ doit()

◆ forciblyReplaceReferencedPayloadOperation()

◆ isValidPackingPermutation()

◆ packMatmulGreedily()

◆ parseMultitileSizesTypes()

◆ parseOptionalInterchange()

◆ printMultitileSizesTypes()

◆ printOptionalInterchange()

◆ replaceForAllWithNewSignature()

◆ tileAndFuseFirstExtractUse()

◆ tileAndFuseFirstExtractUseThroughContainingOpBlockArgument()

◆ tryApply()

◆ unpackSingleIndexResultPayloadOperations() [1/2]

◆ unpackSingleIndexResultPayloadOperations() [2/2]