Abstract
Parallel file systems often work guided by APIs which provide hints to access storage in a coordinated manner. Nevertheless, the current APIs do not offer the expressiveness necessary to specify I/O operations conveniently in irregular applications. We characterize the state of irregular applications that precedes the performance degradation of the parallel file system and we propose a schedule based on profile information. © Springer-Verlag 2006.
Cite
CITATION STYLE
Ibáñez, M. B., García, F., & Carretero, J. (2006). A profiling approach for the management of writing in irregular applications. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 4331 LNCS, pp. 251–259). https://doi.org/10.1007/11942634_27
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
