tvm::s_tir::meta_schedule Namespace Reference

Namespaces
namespace	attr

Classes
class	ArgInfo
	Managed reference to ArgInfoObj. More...
class	ArgInfoObj
	The argument information. More...
class	Builder
	Managed reference to BuilderObj. More...
class	BuilderInput
	Managed reference to BuilderInputObj. More...
class	BuilderInputObj
	The builder's input, containing an IRModule and the target. More...
class	BuilderObj
	The abstract builder interface. More...
class	BuilderResult
	Managed reference to BuilderResultObj. More...
class	BuilderResultObj
	The builder's output, containing the artifact path or error message if any. More...
class	CostModel
	Managed reference to CostModelObj. More...
class	CostModelObj
	Cost model. More...
class	Database
	Managed reference to DatabaseObj. More...
class	DatabaseObj
class	ExtractedTask
	Managed reference to ExtractedTaskObj. More...
class	ExtractedTaskObj
	A tuning task extracted from the high-level IR. More...
class	FeatureExtractor
	Managed reference to FeatureExtractorObj. More...
class	FeatureExtractorObj
	Extractor for features from measure candidates for use in cost model. More...
class	MeasureCallback
	Managed reference to MeasureCallbackObj. More...
class	MeasureCallbackObj
	Rules to apply after measure results is available. More...
class	MeasureCandidate
	Managed reference to MeasureCandidateObj. More...
class	MeasureCandidateObj
	The schedule (with input shapes) to be measured. More...
class	Mutator
	Managed reference to MutatorObj. More...
class	MutatorObj
	Mutator is designed to mutate the trace to explore the design space. More...
class	Postproc
	Managed reference to PostprocObj. More...
class	PostprocObj
	Rules to apply a postprocessor to a schedule. More...
class	Profiler
	Managed reference to ProfilerObj. More...
class	ProfilerObj
	A generic profiler. More...
class	PyBuilderObj
	An abstract builder with customized build method on the python-side. More...
class	PyCostModelObj
	The cost model with customized methods on the python-side. More...
class	PyDatabaseObj
	The database with customized methods on the python-side. More...
class	PyFeatureExtractorObj
	The feature extractor with customized methods on the python-side. More...
class	PyMeasureCallbackObj
	The measure callback with customized methods on the python-side. More...
class	PyMutatorObj
	The mutator with customized methods on the python-side. More...
class	PyPostprocObj
	The postprocessor with customized methods on the python-side. More...
class	PyRunnerObj
	An abstract runner with customized build method on the python-side. More...
class	PyScheduleRuleObj
	The schedule rule with customized methods on the python-side. More...
class	PySearchStrategyObj
	The python side customizable class for measure candidate generation. More...
class	PySpaceGeneratorObj
	The design space generator with customized methods on the python-side. More...
class	PyTaskSchedulerObj
	The task scheduler with customized methods on the python-side. More...
class	Runner
	Managed reference to RunnerObj. More...
class	RunnerFuture
	Managed reference to RunnerFutureObj. More...
class	RunnerFutureObj
	A class to asynchronously fetch runner's output. More...
class	RunnerInput
	Managed reference to RunnerInputObj. More...
class	RunnerInputObj
	Runner's input containing path of artifact, type of device and argument info. More...
class	RunnerObj
	The abstract runner interface. More...
class	RunnerResult
	Managed reference to RunnerResultObj. More...
class	RunnerResultObj
	Runner's output containing measurement result of MeasureCandidate or error msg if any. More...
class	ScheduleRule
	Managed reference to ScheduleRuleObj. More...
class	ScheduleRuleObj
	Rules to modify a block in a schedule. More...
class	ScopedTimer
class	SearchStrategy
	Managed reference to SearchStrategyObj. More...
class	SearchStrategyObj
	The search strategy for measure candidates generation. More...
class	SpaceGenerator
	Managed reference to SpaceGeneratorObj. More...
class	SpaceGeneratorObj
	The abstract class for design space generation. More...
class	TaskRecord
	Managed reference to TaskRecordObj. More...
class	TaskRecordObj
class	TaskScheduler
	Managed reference to TaskSchedulerObj. More...
class	TaskSchedulerObj
	The abstract interface of task schedulers. More...
class	TensorInfo
	Managed reference to TensorInfoObj. More...
class	TensorInfoObj
	The tensor argument information. More...
class	TuneContext
	Managed reference to TuneContextObj. More...
class	TuneContextObj
	The auto tuning context. More...
class	TuningRecord
	The managed reference of TuningRecordObj. More...
class	TuningRecordObj
	The class of tuning records. More...
class	Workload
	Managed reference to WorkloadObj. More...
struct	WorkloadEqual
	The equality check for Workload. More...
struct	WorkloadHash
	The hash method for Workload. More...
class	WorkloadObj
	A workload, i.e. an IRModule and its structural hash. More...

Functions
std::function< s_tir::ExprRV(int64_t)>	MakeFactorSampler (s_tir::Schedule sch, ffi::Array< int64_t > thread_extents)
	Given candidates of thread_extents, make a sampler that use sch->SampleCategorical to return a random thread extent.
ffi::Array< s_tir::LoopRV >	BindSpatialLoop (s_tir::Schedule sch, s_tir::LoopRV loop, int64_t max_threadblocks, int64_t max_threads_per_block, std::function< s_tir::ExprRV(int64_t)> get_factor=nullptr)
	Bind blockIdx.x and threadIdx.x to the given loop.
void	BindBlockThreadIdx (s_tir::Schedule sch, s_tir::SBlockRV block, int64_t max_threadblocks, int64_t max_threads_per_block, std::function< s_tir::ExprRV(int64_t max_extent)> get_factor=nullptr)
	Bind the given block if it is not bound to blockIdx or threadIdx.
s_tir::SBlockRV	GetWinogradProducerAndInlineConst (s_tir::Schedule sch, s_tir::SBlockRV block)
	Get the producer block of a given block. If there is a constant winograd transform matrix, inline it.

Function Documentation

BindBlockThreadIdx()

void tvm::s_tir::meta_schedule::BindBlockThreadIdx	(	s_tir::Schedule	sch,
		s_tir::SBlockRV	block,
		int64_t	max_threadblocks,
		int64_t	max_threads_per_block,
		std::function< s_tir::ExprRV(int64_t max_extent)>	get_factor = nullptr
	)

Bind the given block if it is not bound to blockIdx or threadIdx.

Parameters

sch	The schedule.
block	The block to be bound.
max_threadblocks	The maximum number of threadblocks allowed.
max_threads_per_block	The maximum number of threads allowed.
get_factor	A function that returns the tiling factor.

BindSpatialLoop()

ffi::Array< s_tir::LoopRV > tvm::s_tir::meta_schedule::BindSpatialLoop	(	s_tir::Schedule	sch,
		s_tir::LoopRV	loop,
		int64_t	max_threadblocks,
		int64_t	max_threads_per_block,
		std::function< s_tir::ExprRV(int64_t)>	get_factor = nullptr
	)

Bind blockIdx.x and threadIdx.x to the given loop.

Parameters

sch	The schedule.
loop	The loop to be bound.
max_threadblocks	The maximum number of threadblocks allowed.
max_threads_per_block	The maximum number of threads allowed.
get_factor	A function that returns the tiling factor.

Returns

The binded loops in the order of blockIdx.x, threadIdx.x, and the rest.

GetWinogradProducerAndInlineConst()

s_tir::SBlockRV tvm::s_tir::meta_schedule::GetWinogradProducerAndInlineConst	(	s_tir::Schedule	sch,
		s_tir::SBlockRV	block
	)

Get the producer block of a given block. If there is a constant winograd transform matrix, inline it.

Returns

The only producer block.

MakeFactorSampler()

std::function< s_tir::ExprRV(int64_t)> tvm::s_tir::meta_schedule::MakeFactorSampler	(	s_tir::Schedule	sch,
		ffi::Array< int64_t >	thread_extents
	)

Given candidates of thread_extents, make a sampler that use sch->SampleCategorical to return a random thread extent.

Parameters

sch	The schedule
thread_extents	The candidate thread extents.

Returns

A sampler that returns a random thread extent.