May 8, 1998 U.S. Information Agency 301 4th Street Washington, D.C. 20547 Dear Sir/Madam: This letter is written in support of the request of Dr. Yuhong Wen for the granting of a waiver of his two (2) year foreign residency requirement. Description of DARPA The Defense Advanced Research Projects Agency (DARPA), with its principal office at 3701 North Fairfax Drive, Arlington, VA is the central research and development organization for the Department of Defense (DoD). It manages and directs selected basic and applied research and development projects for DoD, and pursues research and technology where risk and payoff are both very high and where success may provide dramatic advances for traditional military roles and missions and dual-use applications. DARPA's primary responsibility is to help maintain U.S. technological superiority and guard against unforeseen technological advances by potential adversaries. Consequently, the DARPA mission is to develop imaginative, innovative and often high risk research ideas offering a significant technological impact that will go well beyond the normal evolutionary developmental approaches; and, to pursue these ideas from the demonstration of technical feasibility through the development of prototype systems. Description of work performed by Dr. Yuhong Wen Dr. Wen is the technical lead for DARPA's performance estimation tool Petasim for parallel hierarchical memory systems at the Northeast Parallel Architecture Center at Syracuse University. This project is aimed at developing methodologies that will make it possible to provide very rapidly approximate predictions of the performance that could be achieved by sophisticated applications on new high performance architectures. It is a high-performance computing project, which will give critical performance prediction capability when designing and developing future generation computer architecture to reach PetaFlops performance. This will have impact on both new commercial systems as well as those of national security importance. A good example is the Department of Energy's Defense systems ASCI program which requires acceleration in performance of computers to be able to fulfill the agency's "Stewardship of the Nuclear Stockpile" mission where simulation with high performance computers must replace testing of nuclear weapons. Dr. Wen also played a major role in DARPA sponsored research on the construction of a common runtime system support for high performance computing tools. This aims to advance the design and research in the compiler techniques for High Performance FORTRAN, C++ and Java. The deliverables of this project include the development of a public domain common runtime infrastructure supporting parallel compiler/interpreter development and general-purpose high level SPMD programming for distributed- memory machines. Moreover this project pioneered a Java empowered WebTop computing mechanism, which uses this infrastructure as underlying engine, for world wide collective computational solution for Science and engineering. In his current position, Dr. Wen's key responsibility is to design and implement the fundamental performance estimation tool -- Petasim -- which embodies critical new ideas aimed at supporting rapid prototyping of new machine architectures. He has developed an accurate methodology for the abstraction of both the computer architecture and the application execution behavior. This allows Petasim to predict the execution time of an application running on novel architecture parallel systems. Other aspects of DARPA's performance estimation project will be strongly dependent on Dr. Wen's Petasim work. This includes Application Emulators design, and Hierarchical Application Modeling Framework (HLAM) design and research at Maryland University and Rutgers University. Job Responsibilities for Dr. Wen In his capacity as technical lead for the Petasim project, Dr. Wen has responsibility for: 1. Research on the abstraction or modeling of both computer architectures and typical important applications; 2. Design and implementation of the performance estimation system -- Petasim; 3. Collaboration research on the design of Hierarchical Application Modeling Framework (HLAM) which will allow interactive description of new applications. 4. Interaction with other members of the team at Maryland and Rutgers Universities. Description of the Performance Estimation Project: The project goal is to develop methodologies that will make it possible to provide approximate predictions of the performance that could be achieved by sophisticated new applications on new high performance architectures. In particular the Petasim tool will enable rapid experimentation and comparative analysis of a wide range of qualitatively defined architectures at the preliminary design phase. While many of the techniques developed will apply to any type of application, the project focuses on two broad classes of applications. Loosely synchronous adaptive applications include adaptive structured or unstructured multigrid codes, particle methods and multipole codes. Data exploration and data fusion applications are codes that carry out-processing, analysis and exploration of one or several very large data sets. This class includes codes that carry out analysis, exploration and fusion of sensor data from sensors located on different platforms (e.g. satellites, aircraft and ships), and codes that carry out sensor data analysis and fusion of data from conventional high power microscopy, confocal microscopy, electron microscopy, computerized tomography, and magnetic resonance imaging. A very critical job function is the provision and validation of the HLAM/Petasim modeling process and implementing the associated tools for this DARPA performance estimation project. Dr. Wen is uniquely qualified to provide this job function. In this regard, it is imperative that Dr. Wen can stay in U.S. to continue his leadership role in the HLAM/Petasim component of the DARPA performance estimation project. If he is required to leave due to visa issues, it will severely hamper the successful completion of this advanced high performance-computing project. Therefore, we request that USIA grant Dr. Wen a waiver of two-year foreign residence requirement. Qualifications of Dr. Wen The blend of education and practical research experience gives Dr. Wen excellent qualifications for a leadership role in the field of high performance and parallel computing. Dr. Wen obtained a Ph.D. degree in computer science at what is generally regarded as the best such department in China. He has performed research work for many years in parallel programming environments, parallel operating systems, languages, networking, databases and performance evaluation. He has showed both technical talent and leadership abilities. From June 1996 through the present, Dr. Wen has been involved in both the DARPA parallel common runtime consortium for high performance computing and compiler techniques, and the HLAM/Petasim project aimed at performance estimation for new high performance architecture design. This work has been performed at the Northeast Parallel Architecture Center at Syracuse University and these projects are all supported by the Defense Advanced Research Project Agency (DARPA). Summary Dr. Wen's early research and development work in HLAM/Petasim project has showed some very good results in the performance estimation which can be expected to be very helpful in the design of new computer. Dr. Wen's background and experience, and his contributions to date make us believe that he is the suitable person to further this project. We are convinced that he can make great improvement and significant further contributions in this research. Thank you for your consideration in granting the requested waiver so that Dr. Wen may carry out his responsibilities as a research scientist in the United States using his impressive professional background which is otherwise presently unavailable to DARPA. Sincerely,