Performance Evaluation and Application Development using FPGA-enhanced Cloud Computing Nodes

使用 FPGA 增强型云计算节点进行性能评估和应用程序开发

• 批准号: 516136-2017
• 负责人: Han, Jie
• 金额: $ 1.82万
• 依托单位: University of Alberta
• 依托单位国家: 加拿大
• 项目类别: Engage Grants Program
• 财政年份: 2017
• 资助国家: 加拿大
• 起止时间: 2017-01-01 至 2018-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=640679
• 关键词: Performance; Evaluation; Application; Development; using

The cloud computing paradigm provides users with access over the Internet to large clusters of computingnodes . Cloud computing offers several advantages: Primarily, users benefit by only paying for the computingtime that they need without having to purchase their own equipment. In addition, cloud computing clusters areorganized to make it easy for users to scale up their programs to exploit a large number of computing nodesthat operate in parallel and thus speed up computationally demanding calculations. Finally, cloud computingallows users to bid for computer access at less busy times and benefit from lower fees. In recent years thecapabilities of computing nodes have increased from the inclusion of multiple general-purpose centralprocessing units (CPUs). Also, specialized parallel computers with many simple CPUs, called graphicsprocessing units (GPUs) and which were developed originally to accelerate visual effects, have been used tospeed up more general calculations whose parallel structure matches that of the GPU. Cloud computing nodeshave already adopted those advances. More recently, cloud computing nodes have become available whoseperformance is further enhanced by software re-configurable chips, called field-programmable gate arrays(FPGAs). FPGAs provide an implementation technology that is intermediate in performance and cost betweengeneral-purpose software-programmable CPUs (and GPUs) and full-custom chips that only perform specificcalculations very fast. FPGAs have been widely used as platforms for building hardware accelerators where nosuitable full-custom chip is available. However, it is only recently that FGPAs have been provided in cloudcomputing nodes. This project is a collaboration between the principal investigator (PI) and the sponsoringcompany to investigate a series of promising applications of FPGA-enhanced cloud computing nodes forproblems that build upon the PI's expertise and that are likely to open up new business opportunities for thecompany. The proposed Engage grant funded project would allow Eideticom to benefit from the expertise andengineering time of the PI as well as his graduate students at the University of Alberta.

云计算范式为用户提供了通过互联网访问大型计算节点集群的机会。云计算提供了几个优点：首先，用户只需支付他们所需的计算时间，而不必购买自己的设备。此外，云计算集群被组织起来，使用户可以轻松地扩展他们的程序，以利用大量并行操作的计算节点，从而加快计算要求高的计算。最后，云计算允许用户在不那么忙碌的时间竞标计算机访问，并从较低的费用中受益。近年来，由于包含多个通用中央处理单元（CPU），计算节点的能力得到了提高.此外，具有许多简单CPU的专用并行计算机，称为图形处理单元（GPU），最初是为了加速视觉效果而开发的，已被用于加速其并行结构与GPU相匹配的更一般的计算。云计算节点已经采用了这些进步。最近，云计算节点已经变得可用，其性能通过软件可重新配置的芯片（称为现场可编程门阵列（FPGA））进一步增强。FPGA提供了一种实现技术，其性能和成本介于通用软件可编程CPU（和GPU）和仅快速执行特定计算的全定制芯片之间。FPGA已被广泛用作构建硬件加速器的平台，其中没有合适的全定制芯片可用。然而，直到最近才在云计算节点中提供FGPA。该项目是主要研究者（PI）和该公司之间的合作，旨在研究FPGA增强型云计算节点的一系列有前途的应用，以解决建立在PI专业知识基础上的问题，并可能为该公司开辟新的商机。拟议的参与赠款资助的项目将使Eideticom受益于专业知识和工程时间的PI以及他的研究生在阿尔伯塔大学。