Tokenless static data flow using associative templates
Abstract
The static data flow model of computation promises high performance from fine grained parallelism, but conventional token-driven static data flow architectures are inefficient in terms of memory bandwidth and microcycles required per operation. The associative template mechanism, a new application of associative techniques, employs specially configured content-addressable memories to provide efficient flow control for static data flow program execution. It supports static data flow semantics while exhibiting memory bandwidth and microcycle demands comparable to those of conventional uniprocessors. Associative diffusion, a second application of associative methods, provides communication between adjacent nodes of a mesh-connected network of associative-template-based processors. This mechanism achieves nearest-neighbor communication at speeds comparable to intra-node transactions by overlapping domains of associativity across boundaries between neighboring nodes. Together, associative templates and associative diffusion provide a token-free approach to static data flow computation.
References
[1]
Karp, R.M., and Miller, R.E., "'Properties of a Model for Parallel Computations: I~tcrminacy, Termination, Queueing", SIAM Journal of Applied Mathematics, Vol. 14, No. 6, Nov. 1966, pp. 1390-1411.
[2]
Dennis, LB., "Programming Generality, Parallelism and Computer Architecture," Information Processing 68, North-Holland, Ams~rdam, 1969, pp. 484-492.
[3]
Aekerman, W.B., "Data Flow languages," AFIPS Confcrgnce Proceedings, Vol. 48: Proceedings of the 1979 National Computer Conference, pp. 1087-1095.
[4]
Dennis, J.B. and Misunas, D.P., "A Preliminary Architecture for a Basic Data-Flow Processor," Proce.e.dings of the Second Annual Symposium on Computer Architecture, Dec. 1974, pp. 126-132.
[5]
Gurd, J.R., Kirkham, C.C., Watson, I., "The Manchester Prototype Dataflow Computer," Communications of the ACM, January 1985, pp. 34-52.
[6]
Hiraki, K., Shimada, T., Nishida, K., "A Hardware Design of the Sigma-l, a Data Flow Computer for Scientific Computations," Proc~gs of the 1984 International Conference on Parallel Processing, Aug. 1984, pp. 524-531.
[7]
Foster, C.C., Content Addressable Parallel Processors, New York, Van Nostrand Reinhold Co., 1976.
[8]
Smith, A.L, "Cache Memories," ACM Computing Surveys, Vol. 12, No. 3, Sept. 1982, pp. 473-530.
[9]
Archibald, L, Baer, J.-U, "Cache Coherence Protocols: Evaluation Using a Multiprocessor Simulation Model," ACM Transactions on Computer Systems, Vol. 4, No. 4, Nov. 1986, pp. 273-298.
[10]
Stone, H.S., High-Performance Computer Architecture, Reading, Mass.: Addison-Wesley, 1987.
[11]
Sterling, T.L., "Intuitive Templates: A Static Data- Flow Architecture without Tokens," White Paper, Harris Government Systems Sector, May 25, 1987,
[12]
Sterling, T.L., "The YARC Static Data-How Supercomputer," Yarc Note S, Harris Government Systems Sector, Oct. 15, 1985.
[13]
Kung, H.T., "Why Systolic Architectures?" Computer, Vol. 15, No. 1, Jan. 1982, pp. 37-46.
[14]
Mead, C., Conway, L., Introduction to VLS! Systems, Reading, Mass.: Addison-Wesley, 1980.
Index Terms
- Tokenless static data flow using associative templates
Recommendations
A new method of morphological associative memories
ICIC'09: Proceedings of the Intelligent computing 5th international conference on Emerging intelligent computing technology and applicationsThe morphological associative memories (MAM) have many attractive advantages. However, they can not give a guarantee that morphological hetero-associative memories are perfect, even if input patterns are perfect. In addition, the problem with the ...
Improving Flash Wear-Leveling by Proactively Moving Static Data
Motivated by the strong demand for flash memory with enhanced reliability, this work attempts to achieve improved flash-memory endurance without substantially increasing overhead and without excessively modifying popular implementation designs such as ...
Associative memory
Encyclopedia of Computer ScienceData in an associative or content-addressable memory is looked up by properties of its value, rather than by its address as in conventional memory. Such a property might be a pattern of bits at the start of a word or set of words of storage, for example,...
Comments
Please enable JavaScript to view thecomments powered by Disqus.Information & Contributors
Information
Published In
Sponsors
- SIGARCH: ACM Special Interest Group on Computer Architecture
- IEEE-CS: Computer Society
In-Cooperation
- NSF: National Science Foundation
- NASA: National Aeronatics and Space Administration
- Supercomputer Edu & Resch Ctr: Supercomputer Edu & Resch Ctr
Publisher
IEEE Computer Society Press
Washington, DC, United States
Publication History
Published: 01 November 1988
Check for updates
Qualifiers
- Article
Conference
Supercomputing88
Sponsor:
- SIGARCH
- IEEE-CS
Supercomputing88: International Conference for High Performance Computing, Networking, Storage and Analysis
November 12 - 17, 1988
Florida, Orlando, USA
Acceptance Rates
Overall Acceptance Rate 1,516 of 6,373 submissions, 24%
Contributors
Other Metrics
Bibliometrics & Citations
Bibliometrics
Article Metrics
- 0Total Citations
- 227Total Downloads
- Downloads (Last 12 months)10
- Downloads (Last 6 weeks)3
Reflects downloads up to 14 Nov 2024
Other Metrics
Citations
View Options
Get Access
Login options
Check if you have access through your login credentials or your institution to get full access on this article.
Sign in