High-Performance Computing, Parallel Methods and Algorithms in Geophysical Problems
The substantiation of a problem of increase of productivity for high-efficiency complex - Supercomputing is resulted in this report. The basic performance’s criteria are considered for HPC cluster and storage with parallel file system. The analysis of options is carried out for such hardware parameters as algorithm and routeing protocol of data transmission, Infiniband buffer size, Size of io, Size of cpu cache and memory page. The questions of effective use of system software (compiler, open source libraries, batch queuing system, trace and visualization tools) are discussed. The recommendations about adjustment of a high-efficiency complex – Supercomputing, based on HP HPC configuration in NSU Novosibirsk, are made.
Note. Abstracts are published in author's edition
Mail to Webmaster |
[Home] [Conferences] |
© 1996-2008, Siberian Branch of Russian Academy of Sciences, Novosibirsk
Last update: 06-Jul-2012 (11:52:47)