Abstract | ||
---|---|---|
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. |
Year | DOI | Venue |
---|---|---|
2017 | 10.1007/978-3-319-59105-6_2 | COMPUTER INFORMATION SYSTEMS AND INDUSTRIAL MANAGEMENT (CISIM 2017) |
Keywords | Field | DocType |
NVRAM, Distributed cache, Graph processing, Performance optimization | Adjacency matrix,Synchronization,Supercomputer,Computer science,Cache,Parallel computing,Distributed cache,Cache algorithms,Cluster Node,Scalability | Conference |
Volume | ISSN | Citations |
10244 | 0302-9743 | 0 |
PageRank | References | Authors |
0.34 | 9 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Artur Malinowski | 1 | 0 | 0.34 |
Pawel Czarnul | 2 | 121 | 21.11 |