sdfg::fp_math::FuseFMA Class Reference

Contracts a floating-point multiplication feeding an addition into a fused multiply-add. More...

#include <sdfg/fp_math/transformations/fuse_fma.h>

Inheritance diagram for sdfg::fp_math::FuseFMA:

Public Member Functions
	FuseFMA (data_flow::Tasklet &add, bool invalidate_analysis=true)
virtual std::string	name () const override
	Get the name of this transformation.
virtual bool	can_be_applied (builder::StructuredSDFGBuilder &builder, analysis::AnalysisManager &analysis_manager) override
	Check if this transformation can be safely applied.
virtual void	apply (builder::StructuredSDFGBuilder &builder, analysis::AnalysisManager &analysis_manager) override
	Apply this transformation to the SDFG.
virtual void	to_json (nlohmann::json &j) const override
	Serialize this transformation to JSON.
data_flow::Tasklet *	fma () const
	The fused multiply-add tasklet created by apply(), or nullptr before.
Public Member Functions inherited from sdfg::transformations::Transformation
virtual	~Transformation ()=default
virtual void	set_report (PassReportConsumer *report)
	Set the report consumer for transformation feedback.
virtual bool	try_apply (builder::StructuredSDFGBuilder &builder, analysis::AnalysisManager &analysis_manager)
	Attempt to apply the transformation if possible.

Static Public Member Functions
static FuseFMA	from_json (builder::StructuredSDFGBuilder &builder, const nlohmann::json &j)

Additional Inherited Members
Protected Attributes inherited from sdfg::transformations::Transformation
PassReportConsumer *	report_ = nullptr

Detailed Description

Contracts a floating-point multiplication feeding an addition into a fused multiply-add.

Anchored at an fp_add tasklet d = t + c (or d = c + t) whose operand t is a transient scalar that is written only by an fp_mul tasklet t = a * b of the same block and read only by this addition. The pair is replaced by a single fp_fma tasklet d = fma(a, b, c); the product container is removed.

If both operands of the addition are such products, the one on _in1 is contracted. All operands, the product and the result must share the same floating-point precision.

Contraction skips the intermediate rounding of the product, so it is only valid under -ffp-contract=fast semantics.

See also

UnfuseFMA for the inverse.

Constructor & Destructor Documentation

FuseFMA()

sdfg::fp_math::FuseFMA::FuseFMA ( data_flow::Tasklet & add, bool invalidate_analysis = true )

explicit

Member Function Documentation

apply()

virtual void sdfg::fp_math::FuseFMA::apply ( builder::StructuredSDFGBuilder & builder, analysis::AnalysisManager & analysis_manager )

overridevirtual

Apply this transformation to the SDFG.

Parameters

builder	The SDFG builder
analysis_manager	The analysis manager (may be invalidated)

Implements sdfg::transformations::Transformation.

can_be_applied()

virtual bool sdfg::fp_math::FuseFMA::can_be_applied ( builder::StructuredSDFGBuilder & builder, analysis::AnalysisManager & analysis_manager )

overridevirtual

Check if this transformation can be safely applied.

Parameters

builder	The SDFG builder
analysis_manager	The analysis manager providing necessary analyses

Returns

true if all preconditions are met

Implements sdfg::transformations::Transformation.

fma()

data_flow::Tasklet * sdfg::fp_math::FuseFMA::fma

(

)

const

The fused multiply-add tasklet created by apply(), or nullptr before.

from_json()

FuseFMA sdfg::fp_math::FuseFMA::from_json ( builder::StructuredSDFGBuilder & builder, const nlohmann::json & j )

static

name()

virtual std::string sdfg::fp_math::FuseFMA::name ( ) const

overridevirtual

Get the name of this transformation.

Returns

The transformation name (e.g., "LoopInterchange")

Implements sdfg::transformations::Transformation.

to_json()

virtual void sdfg::fp_math::FuseFMA::to_json ( nlohmann::json & j ) const

overridevirtual

Serialize this transformation to JSON.

Parameters

j	JSON object to populate with transformation description

Exceptions

std::logic_error

if not implemented

Implements sdfg::transformations::Transformation.

---

The documentation for this class was generated from the following file:

• opt/include/sdfg/fp_math/transformations/fuse_fma.h