{"title":"A VLSI 128-processor chip for multiresolution image processing","authors":"M. Albanesi, V. Cantoni, M. Ferretti, F. Mainieri","doi":"10.1109/MPCS.1994.367066","DOIUrl":null,"url":null,"abstract":"This paper presents the design of a multiprocessor chip integrating 128 simple processors arranged as a mesh of 8 rows by 16 columns. This mesh of PEs is intermingled with a dual mesh of switching elements, that interconnect the PEs and reconfigure them into one of three topologies. i) an 8-connected standard mesh, ii) a set of 4-connected independent meshes: iii) a quad-tree. This architecture is a viable solution to the problem of embedding into silicon a quad-pyramid, a well known multi-resolution structure for image processing. Local memory is shared between PEs of the same row and the address space increases with the level of the pyramid. The embedding supports fault-tolerance by column substitution.<<ETX>>","PeriodicalId":64175,"journal":{"name":"专用汽车","volume":"55 1","pages":"296-307"},"PeriodicalIF":0.0000,"publicationDate":"1994-05-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"专用汽车","FirstCategoryId":"1087","ListUrlMain":"https://doi.org/10.1109/MPCS.1994.367066","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
This paper presents the design of a multiprocessor chip integrating 128 simple processors arranged as a mesh of 8 rows by 16 columns. This mesh of PEs is intermingled with a dual mesh of switching elements, that interconnect the PEs and reconfigure them into one of three topologies. i) an 8-connected standard mesh, ii) a set of 4-connected independent meshes: iii) a quad-tree. This architecture is a viable solution to the problem of embedding into silicon a quad-pyramid, a well known multi-resolution structure for image processing. Local memory is shared between PEs of the same row and the address space increases with the level of the pyramid. The embedding supports fault-tolerance by column substitution.<>