Generating Efficient Libraries for use in FPGA Resynthesis Algorithms

The ability to efficiently match completelyspecified logic functions to structures of K-input look-up tables (K-LUTs) is a central problem in FPGA resynthesis algorithms. We address the problem of matching completely-specified logic functions of 9 to 12 inputs in K-LUT structures. Our method is based on the off-line generation of libraries of LUT structures. During resynthesis, matching is performed efficiently using NPN encoding and hash table look-ups to find alternative LUT structures for the implementation of logic functions. Generating an effective library of LUT structures may seem prohibitive due to the overwhelming number of potential logic functions which must be considered and represented. We show that, by careful consideration of which logic functions and LUT structures to keep in the library, it is, indeed, possible to generate useful, compact libraries. Numerical results are presented to support our proposed use of precomputed libraries.