78 search results - page 2 / 16 » Analysis of the Component Architecture Overhead in Open MPI |
ICPPW
2005
IEEE
13 years 11 months ago
2005
IEEE
Large–scale parallel applications performing global synchronization may spend a significant amount of execution time waiting for the completion of a barrier operation. Conseque...
IPPS
2007
IEEE
13 years 11 months ago
2007
IEEE
Network performance measurement and prediction is very important to predict the running time of high performance computing applications. The LogP model family has been proven to b...
PPOPP
2006
ACM
13 years 11 months ago
2006
ACM
This work presents a general methodology for estimating the performance of an HPC workload when running on a future hardware architecture. Further, it demonstrates the methodology...
IJHPCN
2006
13 years 5 months ago
2006
As symmetric multiprocessors become commonplace, the interconnection networks and the communication system software in clusters of multiprocessors become critical to achieving high...
DDECS
2007
IEEE
13 years 11 months ago
2007
IEEE
—Current system-on-chip implementations integrate IP blocks from different vendors. Typical problems are incompatibility and integration overheads. This paper presents a case stu...