Sciweavers

Share
ICPP
1996
IEEE

Scheduling of Wavefront Parallelism on Scalable Shared-memory Multiprocessors

10 years 3 months ago
Scheduling of Wavefront Parallelism on Scalable Shared-memory Multiprocessors
Tiling exploits temporal reuse carried by an outer loop of a loop nest to enhance cache locality. Loop skewing is typically required to make tiling legal. This restricts parallelism to wavefronts in the tiled iteration space. For a small number of processors, wavefront parallelism can be efficiently exploited using dynamic selfscheduling with a large tile size. Such a strategy enhances intratile locality, but does not necessarily enhance intertile locality. We show that dynamic self-scheduling performs poorly on scalable shared-memory multiprocessors where smaller tiles are necessary to provide sufficient parallelism--smaller tiles place greater importance on intertile locality. We propose static scheduling strategies which enhance intertile locality for small tiles. Results of experiments on a Convex SPP1000 multiprocessor demonstrate that our strategies outperform dynamic selfscheduling by a factor of up to 2.3 on 30 processors.
Naraig Manjikian, Tarek S. Abdelrahman
Added 07 Aug 2010
Updated 07 Aug 2010
Type Conference
Year 1996
Where ICPP
Authors Naraig Manjikian, Tarek S. Abdelrahman
Comments (0)
books