N. W. Bergmann and J. C. Mudge. Comparing the performance of FPGA-based custom computers with general-purpose computers for DSP applications. In D. A. Buell and K. L. Pocek, editors, Proceedings of IEEE Workshop on FPGAs for Custom Computing Machines, pages 164--171, Napa, CA, April 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... decoder [180] Parallel object recognition, geometric hashing [ Digit recurrence division, square root [100, 99] Various (big num, algebra, 23 etc) 16] Polynomial evaluations [44] On line arithmetic [160] Floating point arithmetic [46] CORDIC [6, 104] Character recognition [164] DSP [30, 13, 118, 93, 106]. Genome sequence matching [94, 98] Engineering, sciences applications [16] 24 1964 1998 FPGA DPGA PRISC FPIC 66 72 68 70 74 78 76 82 80 84 86 88 90 92 96 94 Chimaera GARP GAMA PAM Xilinx XC6200 PRISM SPLASH CPS Estrin s m c TRIPTYCH DISC MATRIX RAW RaPiD CVH PAMBlox ....

N. W. Bergmann and J. C. Mudge. Comparing the performance of FPGA-based custom computers with generalpurpose computers for DSP applications. In D. A. Buell and K. L. Pocek, editors, Proceedings of IEEE Workshop on FPGAs for Custom Computing Machines, pages 164--171, Napa, CA, April 1994.

.... in which these systems excell involve many bit level operations, with notable examples including automatic target recognition [56, 68] cryptography [60, 65] scientific computations [52] genetic algorithms [29, 45] genome sequencing [70] image processing [5, 22, 57, 59, 73] signal processing [9, 17, 18, 27, 41, 51, 55, 58] and artificial neural networks [19, 24, 47] A number of development environments have been constructed, with the Trianus Lola system [28] being of particular interest. It provides designers access to dynamically reconfigurable hardware; however, the Lola language upon which it is based can only ....

N. W. Bergmann and J. C. Mudge, Comparing the Performance of FPGA-Based Custom Computers with General-Purpose Computers for DSP Applications, in Proceedings of IEEE Workshop on FPGAs for Custom Computing Machines, D. A. Buell and K. L. Pocek, eds., Napa, CA, Apr. 1994, pp. 164--171.

....pattern (Figure 12 left) These topologies have the advantage of simplicity, because of the purely local interconnection pattern, as well as easy expandability, since meshes can be grown by adding resources to the edge of the array. Numerous 2D mesh based systems have been built [Kean92, Shaw93, Bergmann94, Blickle94, Hauck94b, Tessier94, Yamada94] as well as 3D meshes [Sample92, Qunot94] Linear arrays, which are essentially 1 dimensional meshes, have also been made [Gokhale90, Filloque93, Raimbault93, Monaghan94] Note that the design of a mesh topology can involve several subtle tradeoffs, with ....

....These chips are usually connected to the FPGAs, and are used as temporary storage for results, as well as general purpose memories for circuit emulation. Other systems have included integer and or floating point ALUs [Wolfe88, Williams91, Lewis93, Benner94, Bakkes96, Knittel96] DSPs [Engels91, Bergmann94, vom Bgel94, Zycad94, Pottinger95] and general purpose processors [Filloque93, Shaw93, Raimbault93, Benner94, Koch94b, vom Bgel94, Zycad94] to handle portions of computations where dedicated chips perform better than FPGA solutions. Another common inclusion into a multi FPGA system is crossbars ....

N. W. Bergmann, J. C. Mudge, "Comparing the Performance of FPGA-Based Custom Computers with General-Purpose Computers for DSP Applications", IEEE Workshop on FPGAs for Custom Computing Machines, pp. 164-171, 1994.

....of two floating point numbers takes 19 clock cycles. Even for some of the high cost chips like the Weitek 4167 a multiply takes 3 clock cycles and an addition 2 clock cycles [13] The lack of powerful, standalone arithmetic units with features like internal registers has been documented previously [14] . Therefore the ADSP21020 was selected due to the fact that it not only can perform all desired operations in a single cycle but can simultaneously do arithmetic operations and memory accesses. By creatively programming the DSP, the DSP chip can be utilized as a high speed FPU under the control ....

Bergmann, N., Mudge, J., "Comparing the Performance of FPGA-Based Custom Computers with General -Purpose Computers for DSP Applications", Proceedings of FPGAs for custom computing machines (1994), pp 164-171.

....would become the dominant factor in the query processing. Data I O [7] most likely will not be the primary bottleneck in future high performance query processing. General Purpose Processors (GPPs) do not provide enough computation power for processing and manipulating such complex data [17,22,23]. It is noticeable that the gap between specialized hardware (ASICs) and GPPs has been broadening; as is exemplified by the appearance of the specialized PC co processing peripherals such as graphics video accelerator cards. Specialization often provides tremendous gains in performance. ....

....a computationintensive routine can be partitioned into FPGA coprocessor(s) and the host processor(s) for parallel execution in order to obtain higher computation throughputs. The applications that have been reported to perform well on FPGAs are stream oriented, signal processing type applications [22,23]. A relatively small algorithm is applied to large regular blocks of data. The data moves through the logic and the results are returned to users as a stream without having to wait for the completion of the evaluation of the entire task. The requirement for a relative small algorithm implies that ....

N.W.Bergmann, J.C. Mudge, "Comparing the Performance of FPGA-based custom computer with generalpurpose computers for DSP applications," Proceedings of IEEE Workshop on FPGAs for Custom Computing Machines, NAPA, CA, April 1994.

....4000 series FPGAs and about a couple of hundred microseconds 9 for 6200 series FPGAs, it is difficult to find applications with a suitable temporal locality of computation. The applications that have been shown to perform well on FPGAs are stream oriented, signal processing type applications [Bergmann94]. A relatively small algorithm is applied to large regular blocks of data. The data moves through the logic. Example applications include digital signal processing (DSP) data encryption, graphics and multimedia. Programming is the basic difficulty that all approaches to exploit parallelism have ....

N. W. Bergmann, J. C. Mudge, Comparing the performance of FPGA-based custom computers with general-purpose computers for DSP applications, Proceedings of IEEE Workshop on FPGAs for Custom Computing Machines, Napa, CA, April 1994.

....processing applications, can be found in [5] 34] and [15] 16] investigates the issues associated with the mapping of DSP algorithms to field programmable technologies but does not focus on final DSP algorithm performance and DSP processor FPGA performance comparison issues. 12] 36] and [8] discuss and attempt to illustrate some of the potential and the advantages of FPGA based DSP systems over ASICs and DSP processors but are each limited in scope. 12] sets forth some of the requirements for high speed DSP and discusses the architecture of a custom field programmable ....

....begins with a short discussion of the potential advantages of FPGAs over ASICs and DSP processors and presents results from FIR IIR filter implementations on Xilinx FPGAs as support. The filter implementations, however, obtain performance only roughly equivalent to DSP processor implementations. [8] presents a rough on paper comparison of FPGA based DSP systems to parallel DSP boards using multiple DSP processors and an DEC Alpha workstation. The authors conclude that FPGA based systems give a modest performance improvement for floating point DSP when using FPGAs coupled with custom ....

[Article contains additional citation context not shown here]

N. W. Bergmann and J. C. Mudge. Comparing the performance of FPGAbased custom computers with general-purpose computers for DSP applications. In D. A. Buell and K. L. Pocek, editors, Proceedings of IEEE Workshop on FPGAs for Custom Computing Machines, pages 164--171, Napa, CA, April 1994.

....in the digital domain. The dotted lines show how the digital stage is being expanded into higher frequencies. Performance is the major advantage of FPGAs over conventional processors.It has been shown that for specific applications FPGAs can achieve speedups over processors of 10 to 100 times[1,2,7,8]. The major advantage of FPGAs over ASICs is programmability, which of course has a performance penalty. However, creating a new configuration on FPGAs means designing a new hardware architecture. Therefore, programming FPGA based coprocessors is an order of magnitude more complicated than ....

N. W. Bergmann, J. C. Mudge, Comparing the performance of FPGA-based custom computers with general-purpose computers for DSP applications, Proceedings of IEEE Workshop on FPGAs for Custom Computing Machines, Napa, CA, April 1994.

....multipliers and multiple memory banks to increase data throughput. The FPGA has also recently generated interest for use in implementing digital signal processing systems due to its ability to implement custom hardware solutions while still maintaining flexibility through device reprogramming [2]. Using the FPGA it is hoped that a significant performance improvement can be obtained over the DSP processor without sacrificing system flexibility. This paper is an attempt to quantify the ability of the FPGA to provide an acceptable performance improvement over the DSP processor in the area of ....

N. W. Bergmann and J. C. Mudge. Comparing the performance of FPGA-based custom computers with general-purpose computers for DSP applications. In D. A. Buell and K. L. Pocek, editors, Proceedings of IEEE Workshop on FPGAs for Custom Computing Machines, pages 164--171, Napa, CA, April 1994.

.... are now a number of commercially available CCM products on the market [31, 32, 33, 34] The approach has spawned a conference series devoted to the application of FPGA based custom computers [13, 14, 15, 16] There has been relatively little work on analyzing the source of speedup provided by CCMs [2, 6, 5]. In many cases, the speedup over conventional workstations has been impressive. For example the SPLASH II machine achieved a performance improvement of about 40,000 over a Sparc 10 workstation on solving a genetic database problem. The SPLASH system also achieved a speedup of about 1000 times a ....

Bergmann, N and Mudge, C. "Comparing the Performance of FPGA-Based Custom Computers with General -Purpose Computers for DSP Applications", IEEE Workshop on FPGAs for Custom Computing Machines, Napa, California, April 1994.

No context found.

N. W. Bergmann and J. C. Mudge. Comparing the performance of FPGA-based custom computers with general-purpose computers for DSP applications. In D. A. Buell and K. L. Pocek, editors, Proceedings of IEEE Workshop on FPGAs for Custom Computing Machines, pages 164--171, Napa, CA, April 1994.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC