Processing

Please wait...

Settings

Settings

Goto Application

1. EP3346390 - WINOGRAD ALGORITHM ON A MATRIX PROCESSING ARCHITECTURE

Office
European Patent Office
Application Number 17207876
Application Date 15.12.2017
Publication Number 3346390
Publication Date 11.07.2018
Publication Kind A1
IPC
G06F 17/15
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
FELECTRIC DIGITAL DATA PROCESSING
17Digital computing or data processing equipment or methods, specially adapted for specific functions
10Complex mathematical operations
15Correlation function computation
G06F 17/14
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
FELECTRIC DIGITAL DATA PROCESSING
17Digital computing or data processing equipment or methods, specially adapted for specific functions
10Complex mathematical operations
14Fourier, Walsh or analogous domain transformations
G06F 17/16
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
FELECTRIC DIGITAL DATA PROCESSING
17Digital computing or data processing equipment or methods, specially adapted for specific functions
10Complex mathematical operations
16Matrix or vector computation
CPC
G06F 17/144
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
FELECTRIC DIGITAL DATA PROCESSING
17Digital computing or data processing equipment or methods, specially adapted for specific functions
10Complex mathematical operations
14Fourier, Walsh or analogous domain transformations ; , e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
141Discrete Fourier transforms
144Prime factor Fourier transforms, e.g. Winograd transforms, number theoretic transforms
G06F 17/153
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
FELECTRIC DIGITAL DATA PROCESSING
17Digital computing or data processing equipment or methods, specially adapted for specific functions
10Complex mathematical operations
15Correlation function computation ; including computation of convolution operations
153Multidimensional correlation or convolution
G06F 17/16
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
FELECTRIC DIGITAL DATA PROCESSING
17Digital computing or data processing equipment or methods, specially adapted for specific functions
10Complex mathematical operations
16Matrix or vector computation ; , e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
G06F 15/80
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
FELECTRIC DIGITAL DATA PROCESSING
15Digital computers in general
76Architectures of general purpose stored program computers
80comprising an array of processing units with common control, e.g. single instruction multiple data processors
Applicants INTEL CORP
Inventors WERNER TONY L
KALAIAH ARAVIND
Designated States
Priority Data 15395542 30.12.2016 US
Title
(DE) WINOGRAD-ALGORITHMUS AUF EINER MATRIXVERARBEITUNGSARCHITEKTUR
(EN) WINOGRAD ALGORITHM ON A MATRIX PROCESSING ARCHITECTURE
(FR) ALGORITHME DE WINOGRAD SUR UNE ARCHITECTURE DE TRAITEMENT DE MATRICE
Abstract
(EN)
In one embodiment, a matrix operation may be performed, wherein the matrix operation comprises a matrix multiplication operation on a plurality of matrix operands. Matrix data may be received from a multi-dimensional memory, wherein the matrix data is associated with the plurality of matrix operands. The plurality of matrix operands may be extracted from the matrix data, wherein the plurality of matrix operands comprises a first matrix operand and a second matrix operand. A first transform may be performed on the first matrix operand to obtain a transformed matrix operand, wherein performing matrix multiplication using the transformed matrix operand is faster than performing matrix multiplication using the first matrix operand. Matrix multiplication may be performed on the transformed matrix operand to obtain a partial result. A second transform may be performed on the partial result to obtain a result of the matrix multiplication operation.

Also published as