
For FullText PDF, please login, if you are a member of IEICE,
or go to Pay Per View on menu list, if you are a nonmember of IEICE.

Reduction of the Number of FPGA Blocks by Maximizing Flexibility of Internal Functions
Takenori KOUDA Shigeru YAMASHITA Yahiko KAMBAYASHI
Publication
IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences
Vol.E81A
No.12
pp.25542562 Publication Date: 1998/12/25
Online ISSN:
DOI:
Print ISSN: 09168508 Type of Manuscript: Special Section PAPER (Special Section on VLSI Design and CAD Algorithms) Category: Logic Synthesis Keyword: logic optimization, FPGAs, SPFDs, PSPFDs,
Full Text: PDF(818.2KB)>>
Summary:
In this paper, we will discuss circuit minimization techniques based on the multiple output capability of FPGA blocks. Since previous methods only consider two independent output functions, we will discuss a more complicated case when the two functions are mutually related. We also discuss a method to maximize flexibility of a specified cell output in the given FPGA block. If a set of possible functions for a cell which will not change the FPGA output function is large, we call that the flexibility of this cell is high. The concept of Sets of Pairs of Functions to be Distinguished (SPFDs) introduced by Yamashita et al. is a powerful tool to minimize a given FPGA circuits. In this paper, an extension of the concept, Priority based SPFDs (PSPFDs) is introduced to maximize the flexibility of output functions realized by such internal cells. By using PSPFDs for our new method, we can utilize the multiple output capability very well. Combination with the previous methods with PSPFDs is also shown to be important. We have implemented these methods and applied them to MCNC benchmarks mapped into 5variable function blocks. To make a comparison with other methods, we have implemented methods using wellknown merging algorithms utilizing the same multiple output capability. Experimental results show that our methods can reduce the number of blocks in the initial circuits by 40% on average. This reduction ratio is 16% higher than that of previous methods.

