Paper Info
Title
High Performance Virtual Machines (HPVM'S): Clusters with Supercomputing API's and Performance
Abstract
The HPVM project provides software which enables high-performance computing on clusters of PCs and workstations using standard supercomputing APIs such as MPI, SHMEM Put/Get, and Global Arrays. HPVMs|High-Performance Virtual Machines|are surprisingly competitive with MPP systems, such as the IBM SP2 and Cray T3D. The Illinois HPVM achieves impressive low-level communication performance across the cluster: one-way latencies of around 11 sec and bandwidths > 50 MBytes/sec|even for small packets (< 256 bytes). Performance at higher levels, such as MPI, is expected to be approximately 17 sec latency and also > 50 MByte/sec bandwidth. Current elements of the HPVM project are: Illinois Fast Messages (FM), MPI-FM, FM-DCS (Dynamic Coscheduling), Put/Get-FM, and Global Arrays-FM. This software is in use at dozens of sites.
Year
Venue
Keywords
1997
PPSC
virtual machine
Field
DocType
Citations 
Cluster (physics),Virtual machine,Supercomputer,Computer science,Parallel computing,Operating system
Conference
34
PageRank 
References 
Authors
7.40
11
7
Name
Order
Citations
PageRank
Andrew A. Chien13696405.97
Scott Pakin21098134.55
Mario Lauria362895.12
matthew s buchanan44811.00
Kay Hane5347.40
Louis A. Giannini65812.02
Jane Prusakova7347.40