Message passing performance of Intel Paragon, IBM SP1 and Cray T3D using PVM
Conference
·
OSTI ID:125619
- Boeing Computer Services, Seattle, WA (United States)
For distributed applications on MPP machinse, message passing performance is a critical factor in the overall speed and scalability of the application. This is particularly true when all-to-all communication between nodes is required. In this study we measured the message passing performance of our PVM implementation of an all-to-all communication method called recursive doubling. The measurements were done on an Intel Paragon, IBM SP1 and CRAY T3D. Using a model of message passing times for recursive doubling, we developed several measures that can be used to compare the message passing performance of MPP machines.
- OSTI ID:
- 125619
- Report Number(s):
- CONF-950212--
- Country of Publication:
- United States
- Language:
- English
Similar Records
ACLMPL: Portable and efficient message passing for MPPs
The ACL Message Passing Library
Implementing a parabolized Navier-Stokes flow solver on the Cray T3D
Conference
·
Tue Sep 19 00:00:00 EDT 1995
·
OSTI ID:202436
The ACL Message Passing Library
Conference
·
Fri Sep 01 00:00:00 EDT 1995
·
OSTI ID:109624
Implementing a parabolized Navier-Stokes flow solver on the Cray T3D
Conference
·
Thu Nov 30 23:00:00 EST 1995
·
OSTI ID:125485