tvm::s_tir::meta_schedule::FeatureExtractor Class Reference

Managed reference to FeatureExtractorObj. More...

#include <feature_extractor.h>

Inheritance diagram for tvm::s_tir::meta_schedule::FeatureExtractor:

Collaboration diagram for tvm::s_tir::meta_schedule::FeatureExtractor:

Public Member Functions
	FeatureExtractor (ffi::UnsafeInit tag, ffi::ObjectPtr< FeatureExtractorObj > data)
	TVM_FFI_DEFINE_OBJECT_REF_METHODS_NOTNULLABLE (FeatureExtractor, ffi::ObjectRef, FeatureExtractorObj)

Static Public Member Functions
static FeatureExtractor	PerStoreFeature (int tensors_per_store=5, int arith_intensity_curve_num_samples=10, int cache_line_bytes=64, bool extract_workload=false)
	Create a feature extractor that extracts features from each TensorStore.
static FeatureExtractor	PyFeatureExtractor (PyFeatureExtractorObj::FExtractFrom f_extract_from)
	Create a feature extractor with customized methods on the python-side.

Detailed Description

Managed reference to FeatureExtractorObj.

See also

FeatureExtractorObj

Constructor & Destructor Documentation

FeatureExtractor()

tvm::s_tir::meta_schedule::FeatureExtractor::FeatureExtractor ( ffi::UnsafeInit  tag, ffi::ObjectPtr< FeatureExtractorObj >  data  )

inlineexplicit

Member Function Documentation

PerStoreFeature()

static FeatureExtractor tvm::s_tir::meta_schedule::FeatureExtractor::PerStoreFeature ( int  tensors_per_store = 5, int  arith_intensity_curve_num_samples = 10, int  cache_line_bytes = 64, bool  extract_workload = false  )

static

Create a feature extractor that extracts features from each TensorStore.

Parameters

tensors_per_store	The number of tensors in each TensorStore; Pad or truncate if necessary.
arith_intensity_curve_num_samples	The number of samples used in the arithmetic intensity curve.
cache_line_bytes	The number of bytes in a cache line.
extract_workload	Whether to extract features in the workload in tuning context or not.

Returns

The feature extractor created.

PyFeatureExtractor()

static FeatureExtractor tvm::s_tir::meta_schedule::FeatureExtractor::PyFeatureExtractor ( PyFeatureExtractorObj::FExtractFrom  f_extract_from )

static

Create a feature extractor with customized methods on the python-side.

Parameters

f_extract_from

The packed function of ExtractFrom.

Returns

The feature extractor created.

TVM_FFI_DEFINE_OBJECT_REF_METHODS_NOTNULLABLE()

tvm::s_tir::meta_schedule::FeatureExtractor::TVM_FFI_DEFINE_OBJECT_REF_METHODS_NOTNULLABLE	(	FeatureExtractor	,
		ffi::ObjectRef	,
		FeatureExtractorObj
	)

---

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

• include/tvm/s_tir/meta_schedule/feature_extractor.h