[go: up one dir, main page]
More Web Proxy on the site http://driver.im/ skip to main content
article

Analysis of FPGA/FPIC switch modules

Published: 01 January 2003 Publication History

Abstract

Switch modules are the most important component of the routing resources in FPGAs/FPICs. Previous works have shown that switch modules with higher routability result in better area performance for practical applications. We consider in this paper an FPGA/FPIC switch-module analysis problem: the inputs consist of a switch-module description and the number of nets required to be routed through the switch module; the question is to determine if there exists a feasible routing for the routing requirements on the switch module. As a fundamental problem for the analysis of switch modules, this problem is applicable to the design and routability evaluation of FPGA/FPIC switch modules and FPGA/FPIC routing. We present a network-flow-based approximation algorithm for this problem. Based on mathematical analyses, we show that this algorithm has provably good performance with the bounds 5 and 5/4 away from the optima for two types of switch modules, respectively. Extensive experiments show that this algorithm is highly accurate and runs very efficiently.

References

[1]
Actel Corp. 1996. FPGA Data Book and Design Guide. Actel Corp., Sunnyvale, CA.
[2]
Aptix, Inc. 1992. FPIC AX1024D. Preliminary Data Sheet, Aptix, Inc., San Jose, CA.
[3]
Betz, V., Rose, J., and A. Marquardt 1999. Architecture and CAD for Deep-Submicron FPGAs. Kluwer Academic Publishers, Boston, MA.
[4]
Betz, V. and Rose, J. 2000. Automatic generation of FPGA rotuing architectures from high-level descriptions. In Proceedings of ACM/SIGDA International Symposium on Field Programmable Gate Arrays (Monterey, CA, Feb.) 175--184.
[5]
Bhat, N. and Hill, D. 1992. Routable technology mapping for LUT FPGAs. In Proceedings of IEEE International Conference on Computer Design, VLSI in Computers and Processors (Cambridge, MA, Oct.). 95--98.
[6]
Brown, S. D., Francis, R. J., Rose, J., and Vranesic, Z. 1992a. Field-Programmable Gate Arrays. Kluwer Academic Publishers, Boston, MA.
[7]
Brown, S. D., Rose, J., and Vranesic, Z. 1992b. A detailed router for field-programmable gate arrays. IEEE Trans. Comput.-Aided Des. Integ. Circ. Syst. 11, 5 (May), 620--627.
[8]
Brown, S. D., Rose, J., and Vranesic, Z. G. 1993. A stochastic model to predict the routability of field-programmable gate arrays. IEEE Trans. Comput.-Aided Des. 12, 12 (Dec.), 1827--1838.
[9]
Chang, Y.-W., Thakur, S., Zhu, K., and Wong, D. F. 1994. A new global routing algorithm for FPGAs. In Proceedings of IEEE/ACM International Conference on Computer-Aided Design (San Jose, CA, Nov.). 356--361.
[10]
Chang, Y.-W., Wong, D. F., and Wong, C. K. 1995a. FPGA global routing based on a new congestion metric. In Proceedings of IEEE International Conference on Computer Design (Austin, TX, Oct.). 372--378.
[11]
Chang, Y.-W., Wong, D. F., and Wong, C. K. 1995b. Design and analysis of FPGA/FPIC switch modules. In Proceedings of IEEE International Conference on Computer Design, VLSI in Computers and Processors (Austin, TX, Oct.). 394--401.
[12]
Chang, Y.-W., Wong, D. F., and Wong, C. K. 1996a. Universal switch modules for FPGA design. ACM Trans. Des. Automat. Electron. Syst. 1, 1 (Jan.), 80--101.
[13]
Chang, Y.-W., Wong, D. F., and Wong, C. K. 1996b. Universal switch-module design for symmetric-array FPGAs. In Proceedings of ACM/SIGDA International Symposium on Field Programmable Gate Arrays (Monterey, CA, Feb.). 80--86.
[14]
Gamal, A. E., Greene, J., Reyneri, J., Rogoyski, E., ElAyat, K. A., and Mohsen, A. 1989. An architecture for electrically configurable gate arrays. IEEE J. Solid-State Circ. 24, 2, 394--398.
[15]
Garey, M. and Johnson, D. 1979. Computers and Intractability: A Guide to the Theory of NP-Completeness. W. H. Freeman, San Francisco, CA.
[16]
Graham, R. L., Knuth, D. E., and Patashnik, O. 1989. Concrete Mathematics. Addison-Wesley, Reading, MA.
[17]
Guo et al. 1992. A 1024 pin universal interconnect array with routing architecture. In Proceedings of IEEE Custom Integrated Circuits Conference (San Diego, CA, April). pp. 4.5.1--4.5.4.
[18]
Hennessy, J. L. and Patterson, D. A. 1996. Computer Architecture: A Quantitative Approach, 2nd ed. Morgan Kaufmann, San Mateo, CA.
[19]
Lemieux, G. and Brown, S. D. 1993. A detailed router for allocating wire segments in field-programmable gate arrays. In Proceedings of ACM/SIGDA Physical Design Workshop (Lake Arrow, CA, April). 215--226.
[20]
Lemieux, G., Brown, S. D., and Vranesic, D. 1997. On two-step routing for FPGAs. In Proceedings of ACM International Symposium on Physical Design (Napa Valley, CA, April 14--16). 60--66.
[21]
Lucent Technologies 1996. Field-Programmable Gate Arrays Data Book. Lucent Technologies, Murray Hill, NJ.
[22]
Marple, D. and Cooke, L. 1992. An MPGA compatible FPGA architecture. In Proceedings of ACM International Symposium on Field-Programmable Gate Arrays (Monterey, CA, Feb.). 39--44.
[23]
Rose, J. and Brown, S. 1991. Flexibility of interconnection structures for field-programmable gate arrays. IEEE J. Solid-State Circ. 26, 3, 277--282.
[24]
Shyu, M., Chang, Y.-D., Wu, G.-M. and Chang, Y.-W. 2000. Generic universal switch blocks. IEEE Trans. Comput. 49, 4 (April), 348-359.
[25]
Sleator, D. D. and Tarjan, R. E. 1983. A data structure for dynamic trees. J. Comput. Syst. Sci. 24, 362--391.
[26]
Thakur, S., Chang, Y.-W., Wong, D. F., and Muthukrishnan, S. 1997. Algorithms for an FPGA switch module routing problem with application to global routing. IEEE Trans. Comput.-Aided Des. Integr. Circ. Syst. 16, 1 (Jan.), 32--46.
[27]
Trimberger, S. 1994. Field-Programmable Gate Array Technologies. Kluwer Academic Publishers, Boston, MA.
[28]
Trimberger, S. and Chene, M. 1992. Placement-based partitioning for lookup-table-based FPGA's. In Proceedings of IEEE International Conference Computer Design, VLSI in Computers and Processors (Cambridge, MA, Oct.). pp. 91--94.
[29]
Wilton, S. 1997. Architecture and Algorithms for Field-Programmable Gate Arrays with embedded Memories. Ph.D. dissertation, University of Toronto, Toronto, Ont. Canada.
[30]
Wu, G.-M. and Chang, Y.-W. 1998. Switch-matrix architecture and routing for FPDs. In Proceedings of ACM International Symposium on Physical Design (Monterey, CA, April 14--16). 158--163.
[31]
Wu, G.-M. and Chang, Y.-W. 1999. Quasi-universal switch matrices for FPD design. IEEE Trans. Comput. 48, 10 (Oct.). 1107--1122.
[32]
Wu, Y.-L., Tsukiyama, S., and Marek-Sadowska, M. 1996. Graph based analysis of 2-D FPGA routing. IEEE Trans. Comput.-Aided Des. Integr. Circ. Syst. 15, 1 (Jan.), 33--44.
[33]
Xilinx, Inc. 1996. The Programmable Logic Data Book. Xilinx, Inc., San Jose, CA.
[34]
Zhu, K., Wong, D. F., and Chang, Y.-W. 1993. Switch module design with application to two-dimensional segmentation design. In Proceedings of IEEE/ACM International Conference on Computer-Aided Design (San Jose, CA, Nov. 7--11). pp. 481--486.

Cited By

View all
  • (2005)Routing architecture optimizations for high-density embedded programmable IP coresIEEE Transactions on Very Large Scale Integration (VLSI) Systems10.1109/TVLSI.2005.85956113:11(1320-1324)Online publication date: 1-Nov-2005
  • (2004)Universal switch blocks for three-dimensional FPGA designIEE Proceedings - Circuits, Devices and Systems10.1049/ip-cds:20040228151:1(49)Online publication date: 2004

Recommendations

Reviews

Partha Sarathi Dasgupta

Field-programmable gate arrays (FPGAs) are useful solutions to cost and time-to-market challenges in the semiconductor industry. This is primarily due to the short design-time turnaround, with low risk and cost, facilitating easily modifiable designs. An FPGA is, in general, an array of logic modules that can be interconnected with the help of vertical and horizontal channels and switch modules. A net can change its routing direction via a switch module. A large circuit that cannot be accommodated in a single FPGA is divided into several parts, with each part being realized by an FPGA, and these FPGAs are in turn embedded and interconnected in a field-programmable interconnect chip (FPIC). Earlier studies revealed that the routability of switch modules in FPGAs has a significant effect in reducing the area requirement. Thus, the study of the routability of switch modules in FPGAs is of great importance in the study of the behavior of FPGAs. This paper addresses such a study. The routability analysis problem studied here is defined by the authors as follows: consider a switch module defined as a W x W block, with W terminals on each side of the block. The switch module has two kinds of switches, namely, crossing (row-column) switches and (row-specific or column-specific) separating switches. It is assumed that connection of any two terminals requires at most one switch being “ON.” The terminals around a switch module can be connected in any one of six different ways, two of which are straight, and four are bent connections. A routing requirement vector (RRV) is also present, which gives the required number of terminal connections in each of the six different ways. Formally, the switch module analysis problem (SMAP) is defined as follows: given the input, a switch module M (switch matrix or switch block) and a RRV, the objective is to find if the given RRV is routable on M . A network-flow-based algorithm is proposed to solve a particular class of SMAP. This is then used in approximating general SMAPs. The network-flow analyzer starts with an input switch module and a RRV, and constructs a network-flow diagram from this, assuming each of the cases of a sink terminal on any one of the four sides, with the other sides being the source. In each of the constructed flow networks, the proposed algorithm attempts to check if there is a feasible flow, where, for each i , s_i supplies a flow n_i (component of given RRV), and the sink t receives a flow of sum of all n_is . This problem is solvable in time O(W(N + W)log W) , where N is the number of switches, and W is the maximum number of terminals associated with any of the six directions. If there does not exist any feasible flow for any value of i , the switch module is not routable for the given RRV; otherwise, pick up another sink side. For the proposed flow analyzer, the performance bounds were analyzed, as measured by the ratio of the sizes of two feasible sets obtained by the flow analyzer and by an ILP-based exact flow analyzer. The theoretical performance bounds obtained for switch matrices and switch blocks are 5 and 5/4 respectively. The authors of the paper then extend the network-flow analyzer to a more general switch-module routing model. The relaxed condition for a connection to be feasible is that it does not contain any doglegs. Hence, in the relaxed model, a connection of two terminals can use up to one crossing switch, and two separating switches. The flow analyzer was implemented in C on a SUN SPARC 5 workstation. Empirical studies were performed for the performance of the flow analyzer on switch matrices, and the correlation between switch-matrix and chip level routability. Run times for computing all routable RRVs for a switch matrix by the flow analyzer and the exact analyzer are also reported. To summarize, the authors show the network-flow-based routability analyzer to be very efficient and accurate. They also claim that proving NP-completeness of SMAP is an unsolved problem. Future work may involve the study of routing capacity for multiple switch modules in series, and the interaction of switch modules and pin-to-track connections. Online Computing Reviews Service

Access critical reviews of Computing literature here

Become a reviewer for Computing Reviews.

Comments

Please enable JavaScript to view thecomments powered by Disqus.

Information & Contributors

Information

Published In

cover image ACM Transactions on Design Automation of Electronic Systems
ACM Transactions on Design Automation of Electronic Systems  Volume 8, Issue 1
January 2003
139 pages
ISSN:1084-4309
EISSN:1557-7309
DOI:10.1145/606603
Issue’s Table of Contents

Publisher

Association for Computing Machinery

New York, NY, United States

Journal Family

Publication History

Published: 01 January 2003
Published in TODAES Volume 8, Issue 1

Permissions

Request permissions for this article.

Check for updates

Author Tags

  1. Computer-aided design of VLSI
  2. FPGA
  3. FPIC
  4. layout
  5. synthesis

Qualifiers

  • Article

Contributors

Other Metrics

Bibliometrics & Citations

Bibliometrics

Article Metrics

  • Downloads (Last 12 months)2
  • Downloads (Last 6 weeks)2
Reflects downloads up to 31 Dec 2024

Other Metrics

Citations

Cited By

View all
  • (2005)Routing architecture optimizations for high-density embedded programmable IP coresIEEE Transactions on Very Large Scale Integration (VLSI) Systems10.1109/TVLSI.2005.85956113:11(1320-1324)Online publication date: 1-Nov-2005
  • (2004)Universal switch blocks for three-dimensional FPGA designIEE Proceedings - Circuits, Devices and Systems10.1049/ip-cds:20040228151:1(49)Online publication date: 2004

View Options

Login options

Full Access

View options

PDF

View or Download as a PDF file.

PDF

eReader

View online with eReader.

eReader

Media

Figures

Other

Tables

Share

Share

Share this Publication link

Share on social media