Table Size Reduction and Optimization in Multiplierless Table-Based Function Evaluation Designs
碩士 === 國立中山大學 === 資訊工程學系研究所 === 101 === In hardware design of the elementary function evaluation, Table-lookup-and-addition method is a category of multiplier-less method, based on several lookup tables and a multi-operand adder to calculate the final function value. Multipartite table method [3] is...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2013
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/31297534248791286907 |
| Summary: | 碩士 === 國立中山大學 === 資訊工程學系研究所 === 101 === In hardware design of the elementary function evaluation, Table-lookup-and-addition method is a category of multiplier-less method, based on several lookup tables and a multi-operand adder to calculate the final function value. Multipartite table method [3] is a popular multiplierless function evaluation design. These multiplierless methods usually have small delay but the table size grows very fast with respect to accuracy .Thus they are only used in applications with low-precision requirements. In this thesis, new table decomposition methods are proposed in order to reduce table size in the multipartite design. The original table is decomposed into two or three smaller tables so that the total table size is efficiently reduced. Experimental results show that the proposed design can significantly reduce the bit count of the tables for different functions compared with the multipartite methods in [3] which is the best table-addition method reported so far.
|
|---|