{"title":"TDD: a technology dependent decomposition algorithm for LUT-based FPGAs","authors":"A. Farrahi, M. Sarrafzadeh","doi":"10.1109/ASIC.1997.617006","DOIUrl":null,"url":null,"abstract":"A major drawback of the previous algorithms that perform decomposition and covering for LUT-based FPGA technology mapping is the lack of a fast, and reasonably accurate evaluation scheme for the decomposition phase. In this paper, we will show how a fast covering algorithm can be used as an evaluation engine for the decomposition phase. We show that decomposition has a significant impact on the quality of the final mapping result. More specifically, we show that starting from the same circuit topology, a blind decomposition leads to mapping results that use an average of 70 to 150% more LUTs compared to the results obtained using a technology driven decomposition algorithm. A technology driven decomposition algorithm is developed based on the proposed idea. Experiments on a number of MCNC benchmark circuits show an average of 12% to 72% improvement on the number of LUTs compared to the previously reported results.","PeriodicalId":300310,"journal":{"name":"Proceedings. Tenth Annual IEEE International ASIC Conference and Exhibit (Cat. No.97TH8334)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1997-09-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings. Tenth Annual IEEE International ASIC Conference and Exhibit (Cat. No.97TH8334)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ASIC.1997.617006","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1
Abstract
A major drawback of the previous algorithms that perform decomposition and covering for LUT-based FPGA technology mapping is the lack of a fast, and reasonably accurate evaluation scheme for the decomposition phase. In this paper, we will show how a fast covering algorithm can be used as an evaluation engine for the decomposition phase. We show that decomposition has a significant impact on the quality of the final mapping result. More specifically, we show that starting from the same circuit topology, a blind decomposition leads to mapping results that use an average of 70 to 150% more LUTs compared to the results obtained using a technology driven decomposition algorithm. A technology driven decomposition algorithm is developed based on the proposed idea. Experiments on a number of MCNC benchmark circuits show an average of 12% to 72% improvement on the number of LUTs compared to the previously reported results.