Kuan-Chung Huang, Yu-Chen Wu, Che-Wei Chang, Tei-Wei Kuo, C. Shih, Qingxu Deng
{"title":"Real-time process synchronization for systems with accelerators","authors":"Kuan-Chung Huang, Yu-Chen Wu, Che-Wei Chang, Tei-Wei Kuo, C. Shih, Qingxu Deng","doi":"10.1145/2663761.2664220","DOIUrl":null,"url":null,"abstract":"This work is motivated by the needs to manage the priority inversion problem without sacrificing the utilization of increasingly popular hardware accelerators. A new mechanism is developed to dedicate accelerators to selected higher-priority tasks. The floor and ceiling priorities of accelerators are thus proposed as an extension of the concept of semaphore priority ceiling to guarantee at most two priority inversions for any real-time task in a uniprocessor system with multiple accelerators. The properties of the proposed concept are explored with respect to blocking behaviors over the CPU and accelerators and are verified by a series of experiments, for which the insight of the simple but effective idea is evaluated and presented.","PeriodicalId":120340,"journal":{"name":"Research in Adaptive and Convergent Systems","volume":"1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2014-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Research in Adaptive and Convergent Systems","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/2663761.2664220","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
This work is motivated by the needs to manage the priority inversion problem without sacrificing the utilization of increasingly popular hardware accelerators. A new mechanism is developed to dedicate accelerators to selected higher-priority tasks. The floor and ceiling priorities of accelerators are thus proposed as an extension of the concept of semaphore priority ceiling to guarantee at most two priority inversions for any real-time task in a uniprocessor system with multiple accelerators. The properties of the proposed concept are explored with respect to blocking behaviors over the CPU and accelerators and are verified by a series of experiments, for which the insight of the simple but effective idea is evaluated and presented.