%0 Conference Proceedings
%T Distributed NVRAM Cache – Optimization and Evaluation with Power of Adjacency Matrix
%+ Gdańsk University of Technology (GUT)
%A Malinowski, Artur
%A Czarnul, Paweł
%Z Part 1: Algorithms
%< avec comité de lecture
%( Lecture Notes in Computer Science
%B 16th IFIP International Conference on Computer Information Systems and Industrial Management (CISIM)
%C Bialystok, Poland
%Y Khalid Saeed
%Y Władysław Homenda
%Y Rituparna Chaki
%I Springer International Publishing
%3 Computer Information Systems and Industrial Management
%V LNCS-10244
%P 15-26
%8 2017-06-16
%D 2017
%R 10.1007/978-3-319-59105-6_2
%K NVRAM
%K Distributed cache
%K Graph processing
%K Performance optimization
%Z Computer Science [cs]
%Z Humanities and Social Sciences/Library and information sciencesConference papers
%X In this paper we build on our previously proposed MPI I/O NVRAM distributed cache for high performance computing. In each cluster node it incorporates NVRAMs which are used as an intermediate cache layer between an application and a file for fast read/write operations supported through wrappers of MPI I/O functions. In this paper we propose optimizations of the solution including handling of write requests with a synchronous mode, additional modes preventing data preloading from a file and synchronization on file close if the solution is used as temporary cache only. Furthermore, we have evaluated the solution for a real application that computes powers of an adjacency matrix of a graph in parallel. We demonstrated superiority of our solution compared to a regular MPI I/O implementation for various powers and numbers of graph nodes. Finally, we presented good scalability of the solution for more than 600 processes running on a large HPC cluster.
%G English
%Z TC 8
%2 https://inria.hal.science/hal-01656233/document
%2 https://inria.hal.science/hal-01656233/file/448933_1_En_2_Chapter.pdf
%L hal-01656233
%U https://inria.hal.science/hal-01656233
%~ SHS
%~ IFIP-LNCS
%~ IFIP
%~ IFIP-TC
%~ IFIP-TC8
%~ IFIP-CISIM
%~ IFIP-LNCS-10244