PARALLEL CALCULATION METHOD AND SYSTEM FOR MULTIPLICATION OF SYMMETRIC MATRIX AND VECTOR

Title:

PARALLEL CALCULATION METHOD AND SYSTEM FOR MULTIPLICATION OF SYMMETRIC MATRIX AND VECTOR

Document Type and Number:

WIPO Patent Application WO/2023/206074

Kind Code:

A1

Abstract:

The present invention provides a parallel calculation method and system for multiplication of a symmetric matrix and a vector. The method comprises: obtaining a symmetric matrix and a vector, and determining the number of threads; performing sub-region division; in each sub-region, defining a first axis and/or a second axis parallel to a symmetry axis; calculating a symmetry axis region and the vector, and writing the symmetry axis region and the vector into a register; performing different first multiplication calculation on the first axis twice to obtain two groups of different data, and respectively writing the two groups of different data into the registers; and performing different second multiplication calculation on the second axis twice to obtain two or four groups of different data, and respectively writing the two or four groups of different data into the registers, so as to a calculation result of multiplication of the symmetric matrix and the vector. According to the solution of the present invention, the multiplication calculation of the symmetric matrix and the vector can be accurately and rapidly realized, and by reasonably planning a processing sequence of the elements in the matrix, resources are reasonably allocated to all the threads, so that writing conflicts can be effectively avoided, multi-thread parallel calculation is realized, the calculation efficiency is high, the load of each thread is balanced, and the thread resource distribution is reasonable.

More Like This:

WO/2011/096690	PORTABLE COMMUNICATION TERMINAL FOR EXTRACTING SUBJECTS OF INTEREST TO THE USER, AND A METHOD THEREFOR
WO/2022/219233	A METHOD, AN APPARATUS AND A COMPUTER PROGRAM PRODUCT FOR NEURAL NETWORK COMPRESSION
JP2014160456	SPARSE VARIABLE OPTIMIZATION DEVICE, SPARSE VARIABLE OPTIMIZATION METHOD, AND SPARSE VARIABLE OPTIMIZATION PROGRAM

Inventors:

LIU CHEN (CN)
WANG HUAMIN (CN)

Application Number:

PCT/CN2022/089298

Publication Date:

November 02, 2023

Filing Date:

April 26, 2022

Export Citation:

Click for automatic bibliography generation Help

Assignee:

ZHEJIANG LINGDI DIGITAL TECH CO LTD (CN)

International Classes:

G06F17/16; G06F9/50

Foreign References:

CN107590106A	2018-01-16
US20090319592A1	2009-12-24
CN109614149A	2019-04-12

Other References:

HONG-YANG ZENG: "The research based on the Epiphany of computing parallel matrix multiplication", ELECTRONIC DESIGN ENGINEERING, vol. 23, no. 24, 20 December 2015 (2015-12-20), pages 52 - 55, XP093104861, DOI: 10.14022/j.cnki.dzsjgc.2015.24.016

Attorney, Agent or Firm:

SHENZHEN IPLUS INTELLECTUAL PROPERTY AGENCY (GENERAL PARTNERSHIP) (CN)

Download PDF:

View/Download PDF PDF Help

Previous Patent: FLAME RETARDANT COPOLYESTER COMPOSITIONS

Next Patent: ELECTROCHEMICAL DUAL-SIGNAL DETECTION METHOD AND DEVICE FOR DETECTING BIOMARKERS