US-China Workshop on Acceleration of High Performance Computing Applications

中美高性能计算应用加速研讨会阿拉斯加费尔班克斯大学北极地区超级计算中心（ARSC）和中国同事将组织一次中美高性能计算应用加速研讨会。本次研讨会将于2010年10月在北京举行，讨论在高性能计算系统上实现高效应用性能的挑战。这些挑战越来越多，主要是由于多核处理器和加速硬件（如gpu和Cell处理器）的普及。中美两国科学家在应对应用加速问题上的共同兴趣已经通过合作研讨会规划和一些面对面的访问得到了证实，但中美两国研究人员在这些主题上的学术交流机会很少。该研讨会将邀请美国加速技术领域的领导者到中国进行为期三天的学术交流与合作，其具体目标是为正在进行的合作研究项目确定双方感兴趣的主题。研讨会将包括教师、学生、初级研究人员和高级研究人员，参与者来自代表性不足的群体。高性能计算中的应用程序加速问题尚未得到解决。一些在新的高性能计算（HPC）系统中使用的相同加速技术在今天也可以找到。台式机和笔记本电脑，包括gpu和多核处理器。然而，HPC系统比HPC系统大几千倍，运行在HPC系统上的程序需要在不同组件之间进行更有效的通信，以及在整个系统中分配计算的方法，还需要解决其他问题，例如处理非常大的输出和单个组件失败的可能性。研讨会将汇集来自两国的专家，分享他们对应用加速技术现状的看法，目标是通过高性能计算加强科学发现和分析。研讨会结束后，中美两国参与者将开展联合研究项目和科学合作。本次研讨会由美国国家科学基金会国际科学与工程办公室和计算机与信息科学与工程理事会的计算与通信基础项目共同资助。

US-China Workshop on High Performance Computing Application AccelerationThe Arctic Region Supercomputing Center (ARSC) at the University of Alaska Fairbanks and Chinese colleagues will organize a US-China workshop on application acceleration for high performance computing (HPC) applications. This workshop, to be held in Beijing in October 2010, will address challenges for achieving efficient application performance on HPC systems. These challenges have grown, mainly due to the prevalence of multicore processors and acceleration hardware such as GPUs and the Cell processor. The common interests of Chinese and American scientists in responding to issues of application acceleration have been confirmed through collaborative workshop planning, and some face-to-face visits, but there are few regular opportunities for scholarly communication between US and China researchers on these topics. The workshop will bring US leaders in acceleration technologies to China for three days of scholarly communication and collaboration, with a specific goal of identifying mutually interesting topics for ongoing collaborative research projects. The workshop will include faculty, students, junior researchers, and senior researchers, with participants from underrepresented groups.Application acceleration in high performance computing is not a solved problem. Some of the same acceleration technologies used in new high performance computing (HPC) systems are found in today?s desktop and notebook computers, including GPUs and multicore processors. However, HPC systems are thousands of times larger, and the programs that run on them need more effective communication among the various components and methods for dividing the computation across the system, and other issues such as dealing with very large output and the potential for individual components to fail need to be addressed. The workshop will bring together experts from the two countries to share their views on the state of the art in application acceleration, with the goal of enhancing scientific discovery and analysis through high performance computing. After the workshop, the US and China participants will pursue joint research projects and scientific collaborations. This workshop is supported by the NSF Office of International Science and Engineering with co-funding from the Computing and Communication Foundations Program of the Computer and Information Science and Engineering Directorate.