Optimizing Non-contiguous Accesses in MPI-IO: Revision history

Jump to navigation Jump to search

Diff selection: Mark the radio buttons of the revisions to compare and hit enter or the button at the bottom.
Legend: (cur) = difference with latest revision, (prev) = difference with preceding revision, m = minor edit.

24 December 2023

  • curprev 14:3714:37, 24 December 2023‎ SatoshiNakamoto talk contribs‎ 3,165 bytes +3,165‎ Created page with "Title: Optimizing Non-contiguous Accesses in MPI-IO Abstract: This research focuses on the optimization of non-contiguous accesses in MPI-IO, a parallel I/O interface. The study classifies different ways of expressing an application's I/O needs in MPI-IO into four levels: level0, level1, level2, and level3. It demonstrates that using level3 requests (non-contiguous, collective) rather than level0 requests (Unix-style) can significantly improve I/O performance. The resea..."
Retrieved from "https://simplesci.org/wiki/Optimizing_Non-contiguous_Accesses_in_MPI-IO"