A scalable implementation of the NAS Parallel Benchmark BT on distributed memory systems

An efficient and scalable implementation of the NAS Parallel Benchmark BT suitable for distributed memory systems, such as the IBM Scalable POWERparallel Systems, is described. The pseudo-application benchmark on the IBM SP systems (SP1 and SP2 with wide nodes) are implemented, and performance resul...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:IBM systems journal Ročník 34; číslo 2; s. 273
Hlavný autor: Naik, Vijay K
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Armonk International Business Machines Corporation 01.01.1995
Predmet:
Aerospace industry
Algorithms
Benchmarks
Case studies
Computer architecture
Computer memory
Parallel processing
US
ISSN:0018-8670
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:An efficient and scalable implementation of the NAS Parallel Benchmark BT suitable for distributed memory systems, such as the IBM Scalable POWERparallel Systems, is described. The pseudo-application benchmark on the IBM SP systems (SP1 and SP2 with wide nodes) are implemented, and performance results on up to 128 processors are presented. The results indicate that the SP architecture delivers good performance on this benchmark, both in terms of raw performance and scalability. To get the level of performance obtained, a combination of techniques was used that included the use of efficient sequential algorithms, the use of scalable partitioning strategies, the use of algorithms to reduce the number of messages, the use of improved data structures to reduce memory requirements and memory references, and some tuning for high cache and register utilization.
Bibliografia:ObjectType-Article-1
SourceType-Scholarly Journals-1
content type line 14
ISSN:0018-8670
DOI:10.1147/sj.342.0273