Statistics for slab caching. More...

#include <SlabCacheStats.hpp>

Public Member Functions
	SlabCacheStats (size_t target_length, size_t non_target_length, size_t target_num_slabs, size_t cache_size_in_elements, bool require_minimum_cache)

	SlabCacheStats (size_t target_length, size_t non_target_length, size_t target_num_slabs, size_t cache_size_in_bytes, size_t element_size, bool require_minimum_cache)

Public Attributes
size_t	slab_size_in_elements

size_t	max_slabs_in_cache

Detailed Description

Statistics for slab caching.

This computes the slab size and the number of slabs to be cached, given the dimensions of the slab and the cache size in bytes. The assumption is that all slabs are of the same shape, partitioning the matrix into regular intervals along the target dimension. Developers should check out CustomDenseChunkedMatrix for some usage examples.

Constructor & Destructor Documentation

◆ SlabCacheStats() [1/2]

tatami_chunked::SlabCacheStats::SlabCacheStats	(	size_t	target_length,
		size_t	non_target_length,
		size_t	target_num_slabs,
		size_t	cache_size_in_elements,
		bool	require_minimum_cache
	)

inline

Parameters

target_length	Length of the target dimension of each slab. For example, if we were iterating through rows of a matrix, `target_length` would be the number of rows spanned by each slab.
non_target_length	Length of the non-target dimension of each slab. For example, if we were iterating through matrix rows, the `non_target_length` would be the number of columns spanned by each slab.
target_num_slabs	Number of slabs required to span the full extent of the target dimension of the matrix. This is used as an upper bound on the value of `max_slabs_in_cache`. For example, if we were iterating through matrix rows, the `target_num_slabs` would be the number of slabs required to span all rows of the matrix.
cache_size_in_elements	Total size of the cache, in terms of the number of data elements.
require_minimum_cache	Whether to enforce a minimum size of the cache for efficient extraction of consecutive dimension elements, even if it exceeds `cache_size_in_elements`.

◆ SlabCacheStats() [2/2]

tatami_chunked::SlabCacheStats::SlabCacheStats	(	size_t	target_length,
		size_t	non_target_length,
		size_t	target_num_slabs,
		size_t	cache_size_in_bytes,
		size_t	element_size,
		bool	require_minimum_cache
	)

inline

Parameters

target_length	Length of the target dimension of each slab. For example, if we were iterating through rows of a matrix, `target_length` would be the number of rows spanned by each slab.
non_target_length	Length of the non-target dimension of each slab. For example, if we were iterating through matrix rows, the `non_target_length` would be the number of columns spanned by each slab.
target_num_slabs	Number of slabs required to span the full extent of the target dimension of the matrix. This is used as an upper bound on the value of `max_slabs_in_cache`. For example, if we were iterating through matrix rows, the `target_num_slabs` would be the number of slabs required to span all rows of the matrix.
cache_size_in_bytes	Total size of the cache, in terms of the number of bytes.
element_size	Size of each data element in the cache, in bytes. This may be zero, e.g., when neither the value nor the index are required during sparse extraction.
require_minimum_cache	Whether to enforce a minimum size of the cache for efficient extraction of consecutive dimension elements, even if it exceeds `cache_size_in_bytes`.

Member Data Documentation

◆ slab_size_in_elements

size_t tatami_chunked::SlabCacheStats::slab_size_in_elements

Size of each slab, in terms of the number of data elements.

◆ max_slabs_in_cache

size_t tatami_chunked::SlabCacheStats::max_slabs_in_cache

Number of slabs that can fit in the cache. This is used as max_slabs in LruSlabCache, OracularSlabCache and friends.

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

tatami_chunked/SlabCacheStats.hpp

Public Member Functions