D. Cesini, A. Ferraro, A. Falabella, F. Giacomini, M. Manzali, U. Marconi, N. Neufeld, S. Valat, B. Voneki
{"title":"在x86低功耗架构上使用InfiniBand进行高吞吐量数据采集,用于LHCb升级","authors":"D. Cesini, A. Ferraro, A. Falabella, F. Giacomini, M. Manzali, U. Marconi, N. Neufeld, S. Valat, B. Voneki","doi":"10.1109/RTC.2016.7543136","DOIUrl":null,"url":null,"abstract":"The LHCb Collaboration is preparing a major upgrade of the detector and the Data Acquisition (DAQ) to be installed during the LHC-LS2. The new Event Builder computing farm for the DAQ requires about 500 nodes, and have to be capable of transporting order of 32 Tbps. The requested performance can possibly be achieved using high-bandwidth data-centre switches and commodity hardware. Several studies are ongoing to evaluate and compare network and hardware technologies, with the aim of optimising the performance and also the purchase and maintenance costs of the system. We are investigating if x86 low-power architectures can achieve equivalent performance as traditional servers when used for multi gigabit DAQ. In this talk we introduce an Event Builder implementation based on InfiniBand network and show preliminary tests with this network technology on x86 low-power architectures, such as Intel Atom C2750 and Intel Xeon D-1540, comparing measured bandwidth and power consumption.","PeriodicalId":383702,"journal":{"name":"2016 IEEE-NPSS Real Time Conference (RT)","volume":"41 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2016-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"High throughput data acquisition with InfiniBand on x86 low-power architectures for the LHCb upgrade\",\"authors\":\"D. Cesini, A. Ferraro, A. Falabella, F. Giacomini, M. Manzali, U. Marconi, N. Neufeld, S. Valat, B. Voneki\",\"doi\":\"10.1109/RTC.2016.7543136\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"The LHCb Collaboration is preparing a major upgrade of the detector and the Data Acquisition (DAQ) to be installed during the LHC-LS2. The new Event Builder computing farm for the DAQ requires about 500 nodes, and have to be capable of transporting order of 32 Tbps. The requested performance can possibly be achieved using high-bandwidth data-centre switches and commodity hardware. Several studies are ongoing to evaluate and compare network and hardware technologies, with the aim of optimising the performance and also the purchase and maintenance costs of the system. We are investigating if x86 low-power architectures can achieve equivalent performance as traditional servers when used for multi gigabit DAQ. In this talk we introduce an Event Builder implementation based on InfiniBand network and show preliminary tests with this network technology on x86 low-power architectures, such as Intel Atom C2750 and Intel Xeon D-1540, comparing measured bandwidth and power consumption.\",\"PeriodicalId\":383702,\"journal\":{\"name\":\"2016 IEEE-NPSS Real Time Conference (RT)\",\"volume\":\"41 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2016-06-06\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2016 IEEE-NPSS Real Time Conference (RT)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/RTC.2016.7543136\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2016 IEEE-NPSS Real Time Conference (RT)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/RTC.2016.7543136","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
摘要
LHC-LS2项目正在准备对探测器和数据采集(DAQ)进行重大升级。用于DAQ的新的Event Builder计算场需要大约500个节点,并且必须能够传输32tbps的数量级。所要求的性能可以使用高带宽数据中心交换机和商用硬件来实现。目前正在进行几项研究,以评估和比较网络和硬件技术,目的是优化性能以及系统的购买和维护成本。我们正在研究x86低功耗架构在用于多千兆数据采集时能否达到与传统服务器相当的性能。在这次演讲中,我们介绍了一个基于InfiniBand网络的Event Builder实现,并展示了该网络技术在x86低功耗架构(如Intel Atom C2750和Intel Xeon D-1540)上的初步测试,比较了测量的带宽和功耗。
High throughput data acquisition with InfiniBand on x86 low-power architectures for the LHCb upgrade
The LHCb Collaboration is preparing a major upgrade of the detector and the Data Acquisition (DAQ) to be installed during the LHC-LS2. The new Event Builder computing farm for the DAQ requires about 500 nodes, and have to be capable of transporting order of 32 Tbps. The requested performance can possibly be achieved using high-bandwidth data-centre switches and commodity hardware. Several studies are ongoing to evaluate and compare network and hardware technologies, with the aim of optimising the performance and also the purchase and maintenance costs of the system. We are investigating if x86 low-power architectures can achieve equivalent performance as traditional servers when used for multi gigabit DAQ. In this talk we introduce an Event Builder implementation based on InfiniBand network and show preliminary tests with this network technology on x86 low-power architectures, such as Intel Atom C2750 and Intel Xeon D-1540, comparing measured bandwidth and power consumption.