{"title":"Automatic high level synthesis of partitioned busses","authors":"Christian Ewering","doi":"10.1109/ICCAD.1990.129909","DOIUrl":null,"url":null,"abstract":"A high level synthesis system maps operations of the behavior specification to functional units, minimizing the number of registers, multiplexers and wires. Arranging the results by a floorplanner often leads to a rather large amount of space for interconnections. This situation can be drastically improved if partitioned busses are generated instead of individual connections. For this purpose, a parameterized and powerful target architecture is defined. A method is presented which partitions a data flow graph towards a bus oriented design. A novel allocation method is introduced for this purpose. First experiments with this approach were successful and led to very small designs.<<ETX>>","PeriodicalId":242666,"journal":{"name":"1990 IEEE International Conference on Computer-Aided Design. Digest of Technical Papers","volume":"113 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1990-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"52","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"1990 IEEE International Conference on Computer-Aided Design. Digest of Technical Papers","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICCAD.1990.129909","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 52
Abstract
A high level synthesis system maps operations of the behavior specification to functional units, minimizing the number of registers, multiplexers and wires. Arranging the results by a floorplanner often leads to a rather large amount of space for interconnections. This situation can be drastically improved if partitioned busses are generated instead of individual connections. For this purpose, a parameterized and powerful target architecture is defined. A method is presented which partitions a data flow graph towards a bus oriented design. A novel allocation method is introduced for this purpose. First experiments with this approach were successful and led to very small designs.<>