Title :
Parallel I/O from the user´s perspective
Author :
Gotwals, Jacob ; Srinivas, Suresh ; Yang, Shelby
Author_Institution :
Dept. of Comput. Sci., Indiana Univ., Bloomington, IN, USA
Abstract :
Parallel I/O systems are gaining popularity as a means for providing scalable high bandwidth I/O. In this paper we develop abstract models of parallel I/O systems and provide empirical results that shout how I/O intensive applications interact with the elements of parallel I/O systems. The abstract models are useful for explaining parallel I/O performance to developers of applications requiring high performance I/O and the experimental results provide insight into the I/O performance of present generation parallel machines. We also identify optimizations for improving I/O performance, which should be useful for developers of applications and I/O libraries
Keywords :
parallel machines; performance evaluation; I/O libraries; abstract models; optimizations; parallel I/O; parallel I/O performance; parallel I/O systems; parallel machines; scalable high bandwidth I/O; user´s perspective; Bandwidth; Computer networks; Computer science; Concurrent computing; Distributed computing; File systems; High performance computing; Jacobian matrices; Libraries; Parallel machines;
Conference_Titel :
Frontiers of Massively Parallel Computation, 1995. Proceedings. Frontiers '95., Fifth Symposium on the
Conference_Location :
McLean, VA
Print_ISBN :
0-8186-6965-9
DOI :
10.1109/FMPC.1995.380455