{"title":"基于NoC网络接口的专用硬件代理的多核soc实时监控","authors":"Georgios Kornaros, D. Pnevmatikatos","doi":"10.1109/IPDPSW.2012.27","DOIUrl":null,"url":null,"abstract":"Network-on-chip based multicore systems need efficient management of a multitude of processing resources, hence avoiding hardware and system software from making inefficient time- and power-decisions at runtime. Hardware event management is a necessary path to assist in high-speed management of captured events and enable efficient reaction mechanisms. This paper proposes different micro architecture alternatives and describes an infrastructure for real-time monitoring and management of network-on-chip based systems. High-speed and energy efficient circuit techniques are deployed for monitoring agents that reside at the network interfaces in order to be configured dynamically and communicate computed statistics to centralized hardware monitor managers of different functionality and complexity. An implementation of a pipelined centralized monitor manager is shown, with the capacity to maintain event ordering and process different types of concurrent events. A single event is served with a latency of seven clock cycles. The presented results of a quantitative evaluation provide guidelines for system-level designers, proving the need for flexible and at the same time efficient filters for real-time monitors inside complex NoC-based SoCs.","PeriodicalId":378335,"journal":{"name":"2012 IEEE 26th International Parallel and Distributed Processing Symposium Workshops & PhD Forum","volume":"38 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2012-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"5","resultStr":"{\"title\":\"Real-Time Monitoring of Multicore SoCs through Specialized Hardware Agents on NoC Network Interfaces\",\"authors\":\"Georgios Kornaros, D. Pnevmatikatos\",\"doi\":\"10.1109/IPDPSW.2012.27\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Network-on-chip based multicore systems need efficient management of a multitude of processing resources, hence avoiding hardware and system software from making inefficient time- and power-decisions at runtime. Hardware event management is a necessary path to assist in high-speed management of captured events and enable efficient reaction mechanisms. This paper proposes different micro architecture alternatives and describes an infrastructure for real-time monitoring and management of network-on-chip based systems. High-speed and energy efficient circuit techniques are deployed for monitoring agents that reside at the network interfaces in order to be configured dynamically and communicate computed statistics to centralized hardware monitor managers of different functionality and complexity. An implementation of a pipelined centralized monitor manager is shown, with the capacity to maintain event ordering and process different types of concurrent events. A single event is served with a latency of seven clock cycles. The presented results of a quantitative evaluation provide guidelines for system-level designers, proving the need for flexible and at the same time efficient filters for real-time monitors inside complex NoC-based SoCs.\",\"PeriodicalId\":378335,\"journal\":{\"name\":\"2012 IEEE 26th International Parallel and Distributed Processing Symposium Workshops & PhD Forum\",\"volume\":\"38 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2012-05-21\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"5\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2012 IEEE 26th International Parallel and Distributed Processing Symposium Workshops & PhD Forum\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/IPDPSW.2012.27\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2012 IEEE 26th International Parallel and Distributed Processing Symposium Workshops & PhD Forum","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/IPDPSW.2012.27","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Real-Time Monitoring of Multicore SoCs through Specialized Hardware Agents on NoC Network Interfaces
Network-on-chip based multicore systems need efficient management of a multitude of processing resources, hence avoiding hardware and system software from making inefficient time- and power-decisions at runtime. Hardware event management is a necessary path to assist in high-speed management of captured events and enable efficient reaction mechanisms. This paper proposes different micro architecture alternatives and describes an infrastructure for real-time monitoring and management of network-on-chip based systems. High-speed and energy efficient circuit techniques are deployed for monitoring agents that reside at the network interfaces in order to be configured dynamically and communicate computed statistics to centralized hardware monitor managers of different functionality and complexity. An implementation of a pipelined centralized monitor manager is shown, with the capacity to maintain event ordering and process different types of concurrent events. A single event is served with a latency of seven clock cycles. The presented results of a quantitative evaluation provide guidelines for system-level designers, proving the need for flexible and at the same time efficient filters for real-time monitors inside complex NoC-based SoCs.