Remote Shared Memory over Sun Fire Link Interconnect
نویسندگان
چکیده
The interconnection networks and the communication system software in clusters of multiprocessors are critical to achieving high performance. Recently, Sun Microsystems has introduced a new system area network, Sun Fire Link interconnect, for its Sun Fire cluster systems with some performance results at the MPI level. Sun Fire Link is a memory-based interconnect with layered system software components that implements a mechanism for user-level messaging based on direct-access to remote memory regions of other nodes. This is referred to as Remote Shared Memory (RSM). This paper assesses the performance of the interconnect at the RSM level, as well as MPI level. We present the performance of different Remote Shared Memory API primitives, and MPI pingpong latency over the Sun Fire Link interconnect on a cluster of Sun Fire 6800s. Our results indicate a better performance for the RSM put primitive than the get primitive for messages of larger than 64 bytes. We also compare the performance of Sun MPI implementation with the RSM level. Keyword System Area Networks, Remote Shared Memory, Clusters of Multiprocessors, Performance Evaluation, MPI
منابع مشابه
Performance Evaluation of the Sun Fire Link SMP Clusters
The interconnection networks and the communication system software are critical in achieving high performance in clusters. Sun Fire Link interconnect, is a memory-based interconnect, where the Sun MPI uses the Remote Shared Memory (RSM) model for its user-level inter-node messaging. This paper presents the overall architecture of the Sun Fire Link and its messaging layer. We provide an in-depth...
متن کاملUltra-high performance communication with MPI and the Sun fireTM link interconnect
We present a new low-latency system area network that provides the ultra-high bandwidth needed to fuse a collection of large SMP servers into a capability cluster. The network adapter exports a remote shared memory (RSM) model that supports low latency kernel bypass messaging. The SunTM MPI library uses the RSM interface to implement a highly efficient memory-to-memory messaging protocol in whi...
متن کاملAn Optical Deeection-routed Multiprocessor Interconnect
We describe the design and implementation of a high-bandwidth packet-switched optical interconnect for a shared-memory multipro-cessor with latency-hiding mechanisms. The network has a direct topology and uses deeection routing to resolve contention, obviating elastic buuers and reducing ight latency. Each node is itself a cache-coherent multiprocessor with local memory, hardware to support mul...
متن کاملDesign and Evaluation of a Switch
Cache coherent non-uniform memory access (CC-NUMA) multiprocessors provide a scal-able design for shared memory but they continue to suuer from large remote memory access latencies due to comparatively slow memory technology and data transfer latencies in the in-terconnection network. In this paper, we propose a novel hardware caching technique, called switch cache, to improve the remote memory...
متن کاملOpenMP Microbenchmarks Version 2.0
Overheads due to synchronisation, loop scheduling and array operations are an important factor in determining the performance of shared memory parallel programs. We present a set of benchmarks to measure these classes of overhead for the language constructs in OpenMP. Results are presented for a Sun Fire 15K, an IBM p690+ and an SGI Altix, each with its own implementation of OpenMP. Significant...
متن کامل