A Practical Methodology to Compress Technology Libraries Using Recursive Polynomial Representation

With the advancement in technology, the libraries used for storing timing, power and other related information of cells have become voluminous. As a result, run-time of loading libraries, that is often governed by I/O or network bottlenecks, has become unacceptably high. Traditionally, this problem is tackled by compressing technology libraries using gzip or other lossless compression technique. In this paper, we propose a practical methodology to compress technology library with a high compression ratio and tolerable errors. The compressed representation is based on representing the discrete functions of the library as suitable polynomials defined recursively with respect to all the independent variables of that function. Our implementation of the proposed methodology shows that the compression ratio of 8-10× can be achieved for realistic error measures.