Abstract
The Concurrent Systems Architecture Group (CSAG) has been investigating high-performance clustering of commodity PCs with gigabit interconnects and a commodity OS (Windows NT). This paper describes several of the major enhancements in the latest release of HPVM 1.9 (High Performance Virtual Machine). The first two of these are concurrent support of multiple gigabit fabrics and shared memory transport, and a self-tuning I/O scheme that manages performance disparities of various PCI implementations. In addition to these innovations, HPVM 1.9 also delivers both lower-latency (8.8 µs) and high-bandwidth (100+ MB/s) messaging. Another major contribution of HPVM 1.9 is the design and implementation a new unified polling strategy that allows simultaneous use of several networks (either VIA or Myrinet) and a shared memory transport without resorting to complex adaptive polling schemes. The implementation of our new high-performance shared memory transport is described in detail. Finally, several system and application benchmark results are given and interpreted to assess the value of these changes.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
American National Standard for Information Systems. Fiber-distributed data interface (FDDI): Token ring media access control (MAC). Technical Report ANSI X3.139-1987, American National Standard for Information Systems, July 1987.
T. M. Anderson and R. S. Cornelius. High-performance switching with fibre channel. In Digest of Papers Compcon 1992, pages 261-268. IEEE Computer Society Press, 1992.
D. Becker, T. Sterling, D. Savarese, J. Dorband, U. Ranawak, and C. Packer. Beowulf: A parallel workstation for scientific computing. In International Parallel Processing Symposium Proceedings, 1995.
N. J. Boden, D. Cohen, R. E. Felderman, A. E. Kulawik, C. L. Seitz, J. N. Seizovic, and W. Su. Myrinet: A gigabit-per-second local-area network. IEEE Micro,15(1):29-36, February 1995.
B. Chun, A. Mainwaring, and D. Culler. Virtual network transport protocols for Myrinet. In Hot Interconnects V Proceedings, August 1997.
K. Connelly and A. A. Chien. FM-QoS: Real-time communication using self-synchronizing schedules. In SC97 Proceedings, November 1997.
P. Druschel, L. L. Peterson, and B. S. Davie. Experiences with a high speed network adaptor: A software perspective. In SIGCOMM '94 Symposium Proceedings, pages 2-13, August 1994.
C. Dubnicki, A. Bilas, Y. Chen, S. Damianakis, and K. Li. VMMC-2: Efficient support for reliable, connection-oriented commnication. In Hot Interconnects V Proceedings, August 1997.
I. Foster, C. Kesselman, R Olson, and S. Tuecke. Nexus: An interoperability layer for parallel and distributed computer systems. Technical Report Version 1.3, Argonne National Laboratory, December 1993.
D. Garcia and W. Watson. Servernet II. In Parallel Computer Routing and Communications Workshop Proceedings, LNCS. Springer-Verlag, 1997.
L. A. Giannini and A. A. Chien. A software architecture for global address space communication on clusters: Put/get on fast messages. In High-Performance Distributed Computing Conference Proceedings, 1998.
F. Hady and B. L. Menezes. The performance of crossbar-based binary hyper-cubes. IEEE Transactions on Computers, 44(10):1208-1215, October 1995.
IEEE. Standard for scalable coherent interface (SCI) specification. Std. 15961992, IEEE, August 1993.
M. Lauria and A. Chien. MPI-FM: High performance MPI on workstation clusters. Journal of Parallel and Distributed Computing,40(1):4-18, January 1997.
M. Lauria, S. Pakin, and A. A. Chien. Efficient layering for high speed communication: Fast messages 2.x. In High-Performance Distributed Computing Conference Proceedings, 1998.
S. Lumetta, A. M. Mainwaring, and D. E. Culler. Multiprotocol active messages on a cluster of SMPs. In SC97 Proceedings, San Jose, CA, November 1997.
S. Pakin, M. Lauria, and A. Chien. High performance messaging on workstations: Illinois fast messages (FM) for Myrinet. In SC95 Proceedings, volume 2, pages 1528-1557, 1995.
L. Prylli and B. Tourancheau. Protocol design for high performance networking: A Myrinet experience. Technical Report 97-22, LIP, Ecole Normale Superieure de Lyon, July 1997.
P. Sobalvarro, S. Pakin, A. Chien, and W. Weihl. FM-DCS: An implementation of dynamic coscheduling on a network of workstations. In ASPLOS-VII NOW/Cluster Workshop, Cambridge, MA, October 1996.
H. Tezuka, A. Hori, and Y. Ishikawa. PM: A high-performance communication library for multi-user parallel environments. Technical Report TR-96-015, Tskuba Research Center, Real World Computing Partnership, November 1996.
. The Virtual Interface Architecture Version 1.0, December 1997. Promoted by Intel, Compaq, and Microsoft and available from <http: //www viarch. org/>.
T. von Eicken, A. Basu, V. Buch, and W. Vogels. U-Net: A user-level network interface for parallel and distributed computing. In 15th ACM Symposium on Operating Systems Principles Proceedings, pages 40-53, December 1995.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2000 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Bruno, G., Chien, A.A., Katz, M.J., Papadopoulos, P.M. (2000). Performance Enhancements for HPVM in Multi-Network and Heterogeneous Hardware. In: Engquist, B., Johnsson, L., Hammill, M., Short, F. (eds) Simulation and Visualization on the Grid. Lecture Notes in Computational Science and Engineering, vol 13. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-57313-2_2
Download citation
DOI: https://doi.org/10.1007/978-3-642-57313-2_2
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-67264-7
Online ISBN: 978-3-642-57313-2
eBook Packages: Springer Book Archive