Abstract
This paper describes an Operating System approach to the problem of delivering low latency high bandwidth communications for PC clusters running a public domain OS like Linux and connected by standard, off-the-shelf networks like Fast-Ethernet. The PARMA 2 project has the main goal of designing the new light-weight protocol suite PRP, in order to drastically reduce the software overhead introduced by TCP/IP. PRP wants to offer at high level a stream socket oriented interface and at low level compatibility with any device driver. High level compatibility is crucial in facilitating the porting on PRP of existing applications or message passing packages. Moreover, an optimized version of MPI, based on PRP and evolution of the widespread MPICH implementation, is under development, allowing for a very effective reduction of the communication latencies in synchronous communications, compared to the TCP/IP-based MPI.
Preview
Unable to display preview. Download preview PDF.
References
G. Bell: 1995 Observations on Supercomputing Alternatives: Did the MPP Bandwagon Lead to a Cul-de-Sac ? Communications of the ACM Vol 30 No. 3 (1996)
L. Colombet and L. Desbat: Speedup and efficiency of large size applications on heterogeneous networks. Proc. EURO-PAR96 (1996)
C. C. Lim and J. P. Ang: Experience on Optimization and Parallelization of Existing Scientific Applications on Network of Workstations. Proc. PDPTA96 (1996)
R. van Drunen, C. van Teylingen and M. Kroontje: The Amfisbaena: A Parallel Supercomputer System Based on i860 as a Generic Platform for Molecular Dynamics Simulations. Proc. PDPTA96 (1996)
M.A.Blumrich, K.Li, R.Alpert, C.Dubnicki, E.W.Felten, and J.Sandberg: Virtual Memory Mapped Network Interface for the SHRIMP Multicomputer. Proc. “International Symposium on Computer Architecture” ISCA94 (1994) 142–153.
T. Sterling, D. Savarese, B. Fryxell, K. Olson, and D. J. Becker: Communication Overhead for Space Science Applications on the Beowulf Parallel Workstation. Proc. “High Performance Distributed Computing” HPDC95 Pentagon City Virginia USA (1995)
H. Lu, S. Dwarkadas, A. L. Cox and W. Zwaenepoel: Message Passing Versus Distributed Shared Memory on Networks of Workstations. Proc. Supercomputing95 (1995)
G. Chiola and G. Ciaccio: GAMMA: a Low-cost Network of Workstations Based on Active Messages. Proc. “5th EUROMICRO workshop on Parallel and Distributed Processing PDP'97” London UK (1997)
W. R. Stevens: Unix Network Programming. Prentice Hall New Jersey (1990)
D. E. Comer and D. L. Stevens: Internetworking with TCP/IP. Prentice Hall New Jersey (1991)
Z. Xu and K. Hwang: Modeling Communication Overhead: MPI and MPL Performance on the IBM SP2. IEEE Parallel & Distributed Technology Vol. 4 No. 1 (1996) 25–42
R. W. Hockney: The Communication Challenge for MPP: Intel Paragon and Meiko CS-2. Parallel Computing Vol. 6 No. 3 (1994) 389–398
J. J. Dongarra and T. Dunigan: Message-Passing Performance of Various Computers. Tec. Report ORNL/TM-13006 Oak Ridge National Laboratory (1996)
P. Marenzoni: Performance Analysis of Cray T3D and Connection Machine CM-5: a Comparison. Proc. Int. Conf. “High-Performance Computing and Networking HPCN95” Milan Italy Springer-Verlag LNCS 919 (1995) 110–117
P. Marenzoni and P. Rossi, Benchmark Kernels as a Tool for Performance Evaluation of MPP's, Concurrency Practice and Experience, 1997, in press, John Wiley & Sons.
G. Destri and P. Marenzoni: Cellular Neural Networks as a General Massively Parallel Computational Paradigm. Special Issue on Cellular Neural Networks of “International Journal of Circuits Theory and Application” Vol. 24 No. 3 (1996) 397–408
R. M. Butler and E. L. Lusk: Monitors, Messages, and Clusters: The p4 Parallel Programming System. Parallel Computing Vol. 20 (1994) 547–564
W. Gropp and E. L. Lusk: MPICH Working Note: Creating a New MPICH Device Using the Channel Interface. Tec. Report Argonne National Laboratory
L.O. Chua and T. Roska: The CNN Paradigm. IEEE Trans. on Circuit and Systems — I Vol. 40 (1993) 147–155
L.O. Chua and L. Yang: Cellular Neural Network: Theory. IEEE Trans. on Circuit and Systems Vol. 35 (1988) 1257–1272
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1997 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Marenzoni, P., Rimassa, G., Vignali, M., Bertozzi, M., Conte, G., Rossi, P. (1997). An Operating System support to low-overhead communications in NOW clusters. In: Panda, D.K., Stunkel, C.B. (eds) Communication and Architectural Support for Network-Based Parallel Computing. CANPC 1997. Lecture Notes in Computer Science, vol 1199. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-62573-9_10
Download citation
DOI: https://doi.org/10.1007/3-540-62573-9_10
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-62573-5
Online ISBN: 978-3-540-68085-7
eBook Packages: Springer Book Archive