PSPVM: Implementing PVM on a High-Speed Interconnect for Worstation Clusters
نویسندگان
چکیده
PSPVM in an implementation of the PVM package on top of ParaStations high-speed interconnent for workstation clusters. The ParaStation system uses user level communication for message exchange and removes the operating system from the critical path of message transmission. ParaStations user interface consists of a user-level socket emulation. Thus, we need only minor changes to the standard PVM package to get it running on the ParaStation system. Throughput of the PSPVM is increased eight times and latency is reduced by a factor of four compared to regular PVM. The remaining latency is mainly (88%) caused by the PVM package itself. The underlying sockets are so fast (25 s) that the PVM package is the limiting factor. PSPVM o ers nearly the raw performance of the network to the user and is object-code compatible to regular PVM. As a consequence, we achieve an application speed-up of four to six over traditional PVM using regular ethernet on a cluster of workstations.
منابع مشابه
PSPVM: Implementing PVM on a high-speed Interconnect for Workstation Clusters
PSPVM in an implementation of the PVM package on top of ParaStations high-speed interconnent for workstation clusters. The ParaStation system uses user level communication for message exchange and removes the operating system from the critical path of message transmission. ParaStations user interface consists of a user-level socket emu-lation. Thus, we need only minor changes to the standard PV...
متن کاملA Common Messaging Layer for MPI and PVM over SCI
This paper describes the design of a common message passing layer for implementing both MPI and PVM over the SCI interconnect in a workstation or PC cluster. The design is focused at obtaining low latency. The message layer encapsulates all necessary knowledge of the underlying interconnect and operating system. Yet, we claim that it can be used to implement such diierent message passing librar...
متن کاملRing Algorithms on Heterogeneous Clusters with PVM: Performance Analysis and Modeling
We analyze the performance obtainable on heterogeneous computing systems with data-parallel ring algorithms for the computation of longand short-range interactions. The algorithms were originally developed for homogeneous parallel systems, where they yield a nearly linear speed-up. Two heterogeneous platforms are considered: a network of ALPHA Unix workstations and a network of Pentium PCs with...
متن کاملParallel Distributed Computing on Sci Workstation Clusters: Early Experiences
Workstation and PC clusters interconnected by SCI (Scalable Coherent Interface) are very promising technologies for high performance cluster computing. Using commercial SBus-SCI interface cards and early system software and drivers, a two-SPARCstation cluster has been constructed for initial testing and evaluation. PVM direct task-to-task data transfers have been adapted to operate on this clus...
متن کاملMapping of coarse-grained applications onto workstation clusters
We present an environment for configuring and coordinating coarse grained parallel applications on workstation clusters. The environment named CoPA is based on PVM and allows an automatic distribution of functional modules as they occur in typical CAE-applications. By implementing link-based communication on top of PVM, CoPA is able to perform a “post-game” analysis of the communication load be...
متن کامل