|
Accession number;99A0182302
|
| Title;Dynamic Granularity Control for Parallel Functional Language on Distributed Memory Machine. |
| Author;
YOSHIIKE HISAO
(Grad. Sch., Keio Univ.)
NAKANISHI MASAKAZU
(Keio Univ., Fac. of Sci. and Technol.)
|
Journal Title;Joho Shori Gakkai Kenkyu Hokoku
|
Journal Code:Z0031B
|
ISSN:0919-6072
|
|
VOL.98;NO.115(HPC-74);PAGE.89-94(1998)
|
| Figure&Table&Reference;FIG.9, TBL.1, REF.8 |
| Pub. Country;Japan |
| Language;Japanese |
| Abstract;In this paper, we propose a queue-machine-based parallel system with dynamic changing length of segment on distributed memory machine "AP1000+". In usual queue-machine-based parallel execution, length of segment is fixed, thus it is hard for efficient performance to set a optimum length of segment before execution. In our approach, we control granularity of task by changing length of segment to get efficient performance independent on applications. In our implementation, length of segment is changed by existence of a task-request-message from other processor and length of local task queue. As a result, our implementation can get performance as same as execution on optimum length of segment in 3 applications we examine. (author abst.) |
|
|
|
Related Articles;
|
