Package org.apache.sysds.runtime.matrix.data

Class LibMatrixNative

java.lang.Object

org.apache.sysds.runtime.matrix.data.LibMatrixNative

public class LibMatrixNative
extends Object

Constructor Summary

Constructors

Constructor	Description
LibMatrixNative()

Method Summary

Modifier and Type	Method	Description
static void	conv2d(MatrixBlock input, MatrixBlock filter, MatrixBlock outputBlock, DnnParameters params)	This method performs convolution (i.e.
static void	conv2dBackwardData(MatrixBlock filter, MatrixBlock dout, MatrixBlock outputBlock, DnnParameters params)	This method computes the backpropagation errors for previous layer of convolution operation
static void	conv2dBackwardFilter(MatrixBlock input, MatrixBlock dout, MatrixBlock outputBlock, DnnParameters params)	This method computes the backpropogation errors for filter of convolution operation
static void	fromFloatBuffer(FloatBuffer buff, double[] output)
static boolean	isMatMultMemoryBound(int m1Rlen, int m1Clen, int m2Clen)
static boolean	isSinglePrecision()
static MatrixBlock	matrixMult(MatrixBlock m1, MatrixBlock m2, MatrixBlock ret, int k)	Performs matrix multiplication using native library if BLAS is available or else falls back to Java BLAS.
static void	tsmm(MatrixBlock m1, MatrixBlock ret, boolean leftTrans, int k)

Methods inherited from class java.lang.Object

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

Constructor Detail

LibMatrixNative

public LibMatrixNative()

Method Detail

isMatMultMemoryBound

public static boolean isMatMultMemoryBound(int m1Rlen,
                                           int m1Clen,
                                           int m2Clen)

matrixMult

public static MatrixBlock matrixMult(MatrixBlock m1,
                                     MatrixBlock m2,
                                     MatrixBlock ret,
                                     int k)

Performs matrix multiplication using native library if BLAS is available or else falls back to Java BLAS.

Parameters:

m1 - lhs matrix block

m2 - rhs matrix block

ret - output matrix block

k - number of threads

Returns:

the ret matrixBlock if allocated otherwise a new matrixBlock.

tsmm

public static void tsmm(MatrixBlock m1,
                        MatrixBlock ret,
                        boolean leftTrans,
                        int k)

conv2d

public static void conv2d(MatrixBlock input,
                          MatrixBlock filter,
                          MatrixBlock outputBlock,
                          DnnParameters params)

This method performs convolution (i.e. cross-correlation) operation on input

Parameters:

input - input batch

filter - filter

outputBlock - output of convolution

params - convolution parameters

conv2dBackwardFilter

public static void conv2dBackwardFilter(MatrixBlock input,
                                        MatrixBlock dout,
                                        MatrixBlock outputBlock,
                                        DnnParameters params)

This method computes the backpropogation errors for filter of convolution operation

Parameters:

input - input image

dout - errors from next layer

outputBlock - output errors

params - convolution parameters

conv2dBackwardData

public static void conv2dBackwardData(MatrixBlock filter,
                                      MatrixBlock dout,
                                      MatrixBlock outputBlock,
                                      DnnParameters params)

This method computes the backpropagation errors for previous layer of convolution operation

Parameters:

filter - filter used in conv2d

dout - errors from next layer

outputBlock - output errors

params - convolution parameters

isSinglePrecision

public static boolean isSinglePrecision()

fromFloatBuffer

public static void fromFloatBuffer(FloatBuffer buff,
                                   double[] output)