Title:
PARALLEL COMPUTING METHOD AND SYSTEM
Document Type and Number:
WIPO Patent Application WO/2021/009901
Kind Code:
A1
Abstract:
The present invention achieves speeding up of parallel computing. This parallel computing method comprises: a step for arranging, into each of a plurality of computing nodes, each piece of first-level small data obtained by dividing data; a step for further dividing the first-level small data into second-level small data in at least one first computing node among the plurality of computing nodes; a step for transferring, in parallel, each piece of the second-level small data from at least one first computing node to the plurality of computing nodes; a step for transferring, in parallel, the transferred second-level small data from each of the plurality of computing nodes to at least one second computing node among the plurality of computing nodes; and a step for reconfiguring the first-level small data by using the second-level small data transferred from the plurality of compute nodes, in at least one second compute node.
Inventors:
URINO YUTAKA (JP)
Application Number:
PCT/JP2019/028252
Publication Date:
January 21, 2021
Filing Date:
July 18, 2019
Export Citation:
Assignee:
PHOTONICS ELECTRONICS TECHNOLOGY RES ASS (JP)
International Classes:
G06F9/50; G06F15/173; G06F17/16
Foreign References:
JP2017201733A | 2017-11-09 | |||
JP2005072812A | 2005-03-17 | |||
JPH11110362A | 1999-04-23 |
Other References:
VAN DE GEIJN R. A., WATTS J.: "SUMMA: scalable universal matrix multiplication algorithm", CONCURRENCY: PRACTICE AND EXPERIENCE, vol. 9, no. 4, 1 April 1997 (1997-04-01), pages 255 - 274, XP055784652, DOI: 10.1002/%28SICI%291096-9128%28199704%299%3A4%3C255%3A%3AAID-CPE250%3E3.0.CO%3B2-2!
Attorney, Agent or Firm:
ONO, Shinjiro et al. (JP)
Download PDF:
Previous Patent: PROTECTION RELAY DEVICE
Next Patent: INFORMATION PROCESSING DEVICE, PROGRAM, AND INFORMATION PROCESSING METHOD
Next Patent: INFORMATION PROCESSING DEVICE, PROGRAM, AND INFORMATION PROCESSING METHOD