CISE Research Instrumentation: Three Projects in Applications and Programming Environments for Cache- Coherent Distributed Memory Machines

96-17530 Singh，Jaswinder 耶利米？奥斯特莱克 普林斯顿大学 CISE研究仪器：高速缓存相关分布式存储器机器的应用和编程环境 仪器拨款将资助一个高速缓存一致的分布式共享内存多处理器。 这些设备将用于计算机科学和工程的几个研究项目，包括：计算天体物理学的并行应用和编程环境。 2.蛋白质结构测定的并行应用与编程环境.系统评估的并行工作负载及其对软件和硬件系统的影响。第一个跨学科项目的重点是开发并行算法和实现计算宇宙学在这个新的，重要的一类架构，并开发集成的编程环境与真实的用户反馈，以促进其使用。 这些环境包括编程语言扩展、运行时系统和性能反馈工具。 第二个项目是与斯坦福大学医学院的Altman和Levitt教授合作的，在不同的应用领域有着类似的重点。 最后一个项目包括将SPLASH基准测试套件（此类系统使用最广泛的套件）扩展到新兴应用程序领域和服务器工作负载。 工作方法将是在这台机器上本地开发应用程序和环境，并在NSF超级计算机中心的大型机器上进行生产模拟和可扩展性研究。本地专用的设备访问显然对研究的各个方面都至关重要。

96-17530 Singh, Jaswinder Ostriker, Jeremiah P Princeton University CISE Research Instrumentation: Applications and Programming Environments for Cache-Coherent Distributed Memory Machines The instrumentation grant will fund a cache-coherent distributed shared memory multiprocessor. The equipment will be used for several research projects in computer science and engineering, including: Parallel Applications and Programming Environments for Computational Astrophysics. 2. Parallel Applications and Programming Environments for Protein Structure Determination.3. Parallel Workloads for System Evaluation, and Their Implications for Software and Hardware Systems. The first interdisciplinary project is focused on developing parallel algorithms and implementations for computational cosmology on this new, important class of architectures, and on developing integrated programming environments-with real user feedback-to facilitate their use. The environments include programming language extensions, run-time systems, and performance feedback tools. The second project, a collaboration with Professors Altman and Levitt at the Stanford University School of Medicine, has a similar focus in a different application domain. The last project includes broadening the SPLASH benchmark suites, the most widely used suite for such systems, into emerging application areas and server workloads. The method of work will be to develop applications and environments locally on this machine, and perform production simulations and scalibility studies on larger machines at NSF supercomputer centers. Local, dedicated access to the equipment is clearly critical to all aspects of the research.