Experience of implementing applicative parallelism on Cray X-MP
The traditional approach on Cray multiprocessing has been multitasking in FORTRAN environments with operating system support. This kind of approach usually requires explicit user intervention to restructure or reformulated the algorithms to control parallel execution. However, it is error-prone and in most cases, only coarse-grain parallelism such as subrountines or functions was explored. In this paper, an implemention of micro-tasking for an applicative language called SISAL on Cray X-MP is reported. The experience indicates where the automatic approach to parallelization works well, as well as sources of inefficient behavior on the Cray X-MP. The experience also suggests some future improvements of SISAL multiprocessing on Cray X-MP. 9 refs., 1 fig.
- Research Organization:
- Lawrence Livermore National Lab., CA (USA)
- DOE Contract Number:
- W-7405-ENG-48
- OSTI ID:
- 7184030
- Report Number(s):
- UCRL-98303; CONF-880975-1; ON: DE88011728
- Resource Relation:
- Conference: CONPAR '88, Manchester, UK, 12 Sep 1988; Other Information: Portions of this document are illegible in microfiche products
- Country of Publication:
- United States
- Language:
- English
Similar Records
Implementing dense linear algebra algorithms using multitasking on the CRAY X-MP-4 (or approaching the gigaflop)
Early MIMD experience with a plasma physics simulation program on the CRAY X-MP