M. Hebert, M. McNeill, B. Shah, R. L. Grimsdale, P. Lister
{"title":"MARTI-A Multiprocessor Architecture for Ray Tracing Images","authors":"M. Hebert, M. McNeill, B. Shah, R. L. Grimsdale, P. Lister","doi":"10.2312/EGGH/EGGH90/069-083","DOIUrl":null,"url":null,"abstract":"Multiprocessor systems are well suited to ray tracing, since each ray can be traced independently. However, the large databases required to model complex scenes create problems of data access. In this paper we propose a multiprocessor architecture for ray tracing which removes the need for duplication of the database at processor level. The database is held on a group processor basis, and resides in shared memory. Many of these groups, or clusters, can be replicated to form a highly parallel multiprocessing system. Results of a software simulation of the architecture are promising, indicating that a large number of processors per cluster is possible.","PeriodicalId":321323,"journal":{"name":"Advances in Computer Graphics Hardware V","volume":"29 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"9","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Advances in Computer Graphics Hardware V","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.2312/EGGH/EGGH90/069-083","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 9
Abstract
Multiprocessor systems are well suited to ray tracing, since each ray can be traced independently. However, the large databases required to model complex scenes create problems of data access. In this paper we propose a multiprocessor architecture for ray tracing which removes the need for duplication of the database at processor level. The database is held on a group processor basis, and resides in shared memory. Many of these groups, or clusters, can be replicated to form a highly parallel multiprocessing system. Results of a software simulation of the architecture are promising, indicating that a large number of processors per cluster is possible.