The cost model with customized methods on the python-side. More...

#include <cost_model.h>

Inheritance diagram for tvm::s_tir::meta_schedule::PyCostModelNode:

Collaboration diagram for tvm::s_tir::meta_schedule::PyCostModelNode:

Public Types
using	FLoad = ffi::TypedFunction< void(ffi::String)>
	Load the cost model from given file location. More...

using	FSave = ffi::TypedFunction< void(ffi::String)>
	Save the cost model to given file location. More...

using	FUpdate = ffi::TypedFunction< void(const TuneContext &, const ffi::Array< MeasureCandidate > &, const ffi::Array< RunnerResult > &)>
	Update the cost model given running results. More...

using	FPredict = ffi::TypedFunction< void(const TuneContext &, const ffi::Array< MeasureCandidate > &, void *p_addr)>
	Predict the running results of given measure candidates. More...

Public Member Functions
void	Load (const ffi::String &path)
	Load the cost model from given file location. More...

void	Save (const ffi::String &path)
	Save the cost model to given file location. More...

void	Update (const TuneContext &context, const ffi::Array< MeasureCandidate > &candidates, const ffi::Array< RunnerResult > &results)
	Update the cost model given running results. More...

std::vector< double >	Predict (const TuneContext &context, const ffi::Array< MeasureCandidate > &candidates)
	Predict the normalized score (the larger the better) of given measure candidates. More...

	TVM_FFI_DECLARE_OBJECT_INFO_FINAL ("s_tir.meta_schedule.PyCostModel", PyCostModelNode, CostModelNode)

Public Member Functions inherited from tvm::s_tir::meta_schedule::CostModelNode
virtual	~CostModelNode ()=default
	Virtual destructor. More...

	TVM_FFI_DECLARE_OBJECT_INFO ("s_tir.meta_schedule.CostModel", CostModelNode, ffi::Object)

Public Attributes
FLoad	f_load
	The packed function to the `Load` function. More...

FSave	f_save
	The packed function to the `Save` function. More...

FUpdate	f_update
	The packed function to the `Update` function. More...

FPredict	f_predict
	The packed function to the `Predict` function. More...

Additional Inherited Members
Static Public Attributes inherited from tvm::s_tir::meta_schedule::CostModelNode
static constexpr const bool	_type_mutable = true

Detailed Description

The cost model with customized methods on the python-side.

Member Typedef Documentation

◆ FLoad

using tvm::s_tir::meta_schedule::PyCostModelNode::FLoad = ffi::TypedFunction<void(ffi::String)>

Load the cost model from given file location.

Parameters

path	The file path.

◆ FPredict

using tvm::s_tir::meta_schedule::PyCostModelNode::FPredict = ffi::TypedFunction<void(const TuneContext&, const ffi::Array<MeasureCandidate>&, void* p_addr)>

Predict the running results of given measure candidates.

Parameters

context	The tuning context.
candidates	The measure candidates.
p_addr	The address to save the estimated running results.

◆ FSave

using tvm::s_tir::meta_schedule::PyCostModelNode::FSave = ffi::TypedFunction<void(ffi::String)>

Save the cost model to given file location.

Parameters

path	The file path.

◆ FUpdate

using tvm::s_tir::meta_schedule::PyCostModelNode::FUpdate = ffi::TypedFunction<void(const TuneContext&, const ffi::Array<MeasureCandidate>&, const ffi::Array<RunnerResult>&)>

Update the cost model given running results.

Parameters

context	The tuning context.
candidates	The measure candidates.
results	The running results of the measure candidates.

Returns: Whether cost model was updated successfully.

Member Function Documentation

◆ Load()

void tvm::s_tir::meta_schedule::PyCostModelNode::Load ( const ffi::String & path )

virtual

Load the cost model from given file location.

Parameters

path	The file path.

Implements tvm::s_tir::meta_schedule::CostModelNode.

◆ Predict()

std::vector<double> tvm::s_tir::meta_schedule::PyCostModelNode::Predict	(	const TuneContext &	context,
		const ffi::Array< MeasureCandidate > &	candidates
	)

virtual

Predict the normalized score (the larger the better) of given measure candidates.

Parameters

context	The tuning context.
candidates	The measure candidates.

Returns: The predicted normalized score.

Implements tvm::s_tir::meta_schedule::CostModelNode.

◆ Save()

void tvm::s_tir::meta_schedule::PyCostModelNode::Save ( const ffi::String & path )

virtual

Save the cost model to given file location.

Parameters

path	The file path.

Implements tvm::s_tir::meta_schedule::CostModelNode.

◆ TVM_FFI_DECLARE_OBJECT_INFO_FINAL()

tvm::s_tir::meta_schedule::PyCostModelNode::TVM_FFI_DECLARE_OBJECT_INFO_FINAL	(	"s_tir.meta_schedule.PyCostModel"	,
		PyCostModelNode	,
		CostModelNode
	)

◆ Update()

void tvm::s_tir::meta_schedule::PyCostModelNode::Update	(	const TuneContext &	context,
		const ffi::Array< MeasureCandidate > &	candidates,
		const ffi::Array< RunnerResult > &	results
	)

virtual

Update the cost model given running results.

Parameters

context	The tuning context.
candidates	The measure candidates.
results	The running results of the measure candidates.

Implements tvm::s_tir::meta_schedule::CostModelNode.

Member Data Documentation

◆ f_load

FLoad tvm::s_tir::meta_schedule::PyCostModelNode::f_load

The packed function to the Load function.

◆ f_predict

FPredict tvm::s_tir::meta_schedule::PyCostModelNode::f_predict

The packed function to the Predict function.

◆ f_save

FSave tvm::s_tir::meta_schedule::PyCostModelNode::f_save

The packed function to the Save function.

◆ f_update

FUpdate tvm::s_tir::meta_schedule::PyCostModelNode::f_update

The packed function to the Update function.

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

include/tvm/s_tir/meta_schedule/cost_model.h

Public Types

Public Member Functions

Public Attributes

Additional Inherited Members

Detailed Description

Member Typedef Documentation

◆ FLoad

◆ FPredict

◆ FSave

◆ FUpdate

Member Function Documentation

◆ Load()

◆ Predict()

◆ Save()

◆ TVM_FFI_DECLARE_OBJECT_INFO_FINAL()

◆ Update()

Member Data Documentation

◆ f_load

◆ f_predict

◆ f_save

◆ f_update