org.apache.sysds.hops.estim

Class EstimatorBitsetMM

java.lang.Object

org.apache.sysds.hops.estim.SparsityEstimator

org.apache.sysds.hops.estim.EstimatorBitsetMM

public class EstimatorBitsetMM
extends SparsityEstimator

This estimator implements a naive but rather common approach of boolean matrix multiplies which allows to infer the exact non-zero structure and thus is also useful for sparse result preallocation. For example, the following paper indicates that this approach is used for sparse spGEMM in NVIDIA cuSPARSE and Intel MKL: Weifeng Liu and Brian Vinter. An Efficient GPU General Sparse Matrix-Matrix Multiplication for Irregular Data. In IPDPS, pages 370–381, 2014.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	EstimatorBitsetMM.BitsetMatrix
static class	EstimatorBitsetMM.BitsetMatrix1 This class represents a boolean matrix and provides key operations.
static class	EstimatorBitsetMM.BitsetMatrix2

Nested classes/interfaces inherited from class org.apache.sysds.hops.estim.SparsityEstimator

SparsityEstimator.OpCode

Field Summary

Fields inherited from class org.apache.sysds.hops.estim.SparsityEstimator

MIN_PAR_THRESHOLD, MULTI_THREADED_BUILD, MULTI_THREADED_ESTIM

Constructor Summary

Constructors

Constructor and Description
EstimatorBitsetMM()

Method Summary

Modifier and Type	Method and Description
static EstimatorBitsetMM.BitsetMatrix	createBitset(int m, int n)
static EstimatorBitsetMM.BitsetMatrix	createBitset(MatrixBlock in)
double	estim(MatrixBlock m1, MatrixBlock m2) Estimates the output sparsity for a single matrix multiplication.
double	estim(MatrixBlock m1, MatrixBlock m2, SparsityEstimator.OpCode op) Estimates the output sparsity for a given binary operation.
double	estim(MatrixBlock m, SparsityEstimator.OpCode op) Estimates the output sparsity for a given unary operation.
DataCharacteristics	estim(MMNode root) Estimates the output sparsity of a DAG of matrix multiplications for the given operator graph of a single root node.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

EstimatorBitsetMM

public EstimatorBitsetMM()

Method Detail

estim

public DataCharacteristics estim(MMNode root)

Description copied from class: SparsityEstimator

Estimates the output sparsity of a DAG of matrix multiplications for the given operator graph of a single root node.

Specified by:

estim in class SparsityEstimator

Parameters:

root - DAG root node

Returns:

output data characteristics

estim

public double estim(MatrixBlock m1,
                    MatrixBlock m2)

Description copied from class: SparsityEstimator

Estimates the output sparsity for a single matrix multiplication.

Specified by:

estim in class SparsityEstimator

Parameters:

m1 - left-hand-side operand

m2 - right-hand-side operand

Returns:

sparsity

estim

public double estim(MatrixBlock m1,
                    MatrixBlock m2,
                    SparsityEstimator.OpCode op)

Description copied from class: SparsityEstimator

Estimates the output sparsity for a given binary operation.

Specified by:

estim in class SparsityEstimator

Parameters:

m1 - left-hand-side operand

m2 - right-hand-side operand

op - operator code

Returns:

sparsity

estim

public double estim(MatrixBlock m,
                    SparsityEstimator.OpCode op)

Description copied from class: SparsityEstimator

Estimates the output sparsity for a given unary operation.

Specified by:

estim in class SparsityEstimator

Parameters:

m - left-hand-side operand

op - operator code

Returns:

sparsity

createBitset

public static EstimatorBitsetMM.BitsetMatrix createBitset(int m,
                                                          int n)

createBitset

public static EstimatorBitsetMM.BitsetMatrix createBitset(MatrixBlock in)