Abstract
We describe the design of a communications board that supports collective communications operations of parallel programs. Each cluster of processors are interconnected by a bus and connected to the communications board. The communications boards are, in turn, connected to a low latency, high bandwidth, slow configuration time crossbar network. The design is geared towards low-end (inexpensive) parallel computer architectures.
Such an architecture is motivated by recent advances in parallel programming models and in particular by the “loosely synchronous” style of parallel programming. Subgroups of processors alternatively perform local work with collective communication operations. This allows time to reconfigure the large crossbar and enables communication between processors without interrupts. The important broadcast, scatter, gather, and transpose operations can be efficiently supported.
This work has been supported in part by a US-Israel BSF Grant #88-45 and France-Israel BSF Grant #3310
This is a preview of subscription content, log in via an institution to check access.
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
A. Barak and R. Ben-Natan, “Assignment of Tasks to Parallel Architectures,” Technical Report 92-01, Department of Computer Science, The Hebrew University, Jerusalem, Israel, 1992.
K. Hwang, M. Dubois, D.K. Panda, S. Rao, S. Shang, A. Uresin, W. Mao, H. Nair, M. Lytwyn, F. Hsieh, J. Liu, S. Mehrotra, and C.M. Cheng, “OMP: A RISC-based Multiprocessor using Orthogonal-Access Memories and Multiple Spanning Buses,” 1990 International Conference on Supercomputing, Amsterdam, The Netherlands, September 1990, pp. 7–22.
V. Bala, M. C. Chen, S. Kipnis, C-Y. Lin, L. Rudolph, and M. Snir, “A high-level communication library for massively parallel distributed-memory machines,” Internal Memorandum, IBM T. J. Watson Research Center.
G. Fox, M. Johnson, G. Lyzenga, S. Otto, J. Salmon, and D. Walker, Solving Problems on Concurrent Processors, Volume I: General Techniques and Regular Problems, Prentice-Hall, Englewood Cliffs, New Jersey, 1988.
G. Lerman and L. Rudolph, Parallel Processors: Will They Ever Meet? A Three Decade Survey of Parallel Computers.
L. Rudolph and E. Shenfeld, “Interconnection Cache for Large Interconnection Networks,” Manuscript, Department of Computer Science, Hebrew University, Jerusalem, Israel, 1992.
E. Shenfeld, “A Parallel Architecture for a Digital Optical Computer,” Ph.D. Thesis, Department of Computer Science, The Hebrew University, Jerusalem, Israel, 1992.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1993 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Rudolph, L. (1993). Hardware support for collective communication operations. In: Meyer, F., Monien, B., Rosenberg, A.L. (eds) Parallel Architectures and Their Efficient Use. Nixdorf 1992. Lecture Notes in Computer Science, vol 678. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-56731-3_11
Download citation
DOI: https://doi.org/10.1007/3-540-56731-3_11
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-56731-8
Online ISBN: 978-3-540-47637-5
eBook Packages: Springer Book Archive