Responses to RWCP Research Results

-From video interviews -

These video programs include interviews with visitors to RWCP demonstrations at some international conferences overseas such as SC98 and AAAI.

This special edition featurres some of those interviews to inform RWC NEWS readers of how the RWC project impressed people concerned both in Japan and abroad.

"Toward Seamless Computing-RWCP Research Results in SC98" produced by Parallel and Distributed System Software TRC Laboratory SC98 is the world's largest international conference in the field of parallel and distributed computing. We have been participating in the research exhibition of SC since 1996 and gave demonstratons. At SC98, a distributed laboratory of our parallel and distributed computing field made several presentation programs using "RWC PC Cluster II" consisting of 128 PCs and "RWC Alpha Cluster I." The following are comments from two experts interviewed at the SC98 venue."

Interview with Prof. Andrew Grimshaw

"Clusters are seemed as very important cost effective way of doing business now, in this country at least. Increasingly, cluster computers put forward in regular base to replace traditional supercomputers. One of the things I was always impressed with RWC computing cluster was that it was actually manufactured to be a cluster and they actually did a work to do nice packaging, make it altogether in one place and really good, much perfection about it. A lot of other clusters have been more of the "Let's just stock up some computers and see what's going to happen!" And I think they have done a very good job with this machine. This is a significant computational resource, and we would like it to be a part of our network. I want to stress that when it become a part of network, we are not taking over the machine. It is more like storing the community of machines and so it can be moved around in between. And so we are really interested in getting this machine into our network, because it's such a, It's a large machine." "They are one of the sets of people that, sort of pushing that frontier of particularly Linux, Extreme Linux sort of stuff, keeping staff going. I think it's a good job, I really think they are doing nice job."

Andrew S. Grimshaw is an Associate Professor of Computer Science and director of the Institute of Parallel Computation at the University of Virginia. He is the chief designer and architect of Mentat and Legion.

Interview with Prof. Dennis Gannon

"The MPC++ software system which has had a big influence on my own work. I think the work has been done on applications. It's going to be very important as part of our international partnership with the international grid in the transpac..., Asian Pacific to US high performance networking. Because it allows scientists in the United States direct access to the good tools and applications developed by RWCP." "The RWCP architecture for a PC Cluster is the best PC Cluster architecture, I think in the world. The most projects only come close to the potential of what RWCP has done with the PC cluster related research including the software."

Dr. Dennis Gannon is a professor and the chair of Computer Science at Indiana University, and the director of the High-Performance Computing Software Laboratory.

Interview with Dr. John Reynders

"Most of the systems now, typically trying to run in a homogeneous environment where all the components are the same. The challenge, however, is as you trying to grow a system, the components you might get six month from now will be different, have different properties from the components you are getting in year from now, or half year from now. And what you would like to do is to make sure that as you grow the system, you can build up on everything that already in place rather than swapping out everything and putting in something new. The challenge there rise to build the software structure is they can hide the complexity in the heterogeneous environment. So, the application developer looks like, he or she is just programming one system, rather than collection of several systems. I think that's going to be a big challenge facing software developers for high performance computing over the next 5 years how to make heterogeneous systems easy and transparent to use." "I think our collaboration with RWCP will greatly help our ability to understand software inter structure challenges of encapsulating complexity of such a system. We have enjoyed collaboration of passing years. We have learned a lot from them RWCP folks, and I think they have learned a lot from us, and we view this as technical interchange interactions."

Dr. John Reynders is Director of the Advanced Computing Laboratory and is the Senior Project leader of computational sciences for the ASCI program at the U.S. Los Alamos National Laboratory.

Interview with Dr. Peter Beckman

"The SCore system provides us a way to organize our cluster in a coherent way. Without that system we will be lost for how to schedule, and get all of the jobs running at same time, and getting at communicating at same time, and doing Gang scheduling. And it has been a very important middle ware layer between application and down at the low-level hardware. So, those layers of software have been critical for us to build on that staff instead of reinvent our own system, we have been building on software that RWC has produced and layering our software on top of that." "We were gravitated to using SCore system because of its one of the advanced layers of this sort that covers both SMP and scheduling and with completely integrated package. And I think an important next step is to get this out to a larger community, so it can become de facto standard, so it can become layer that people can program to. The next step will be to make that available to wider audience and have many people using the software and then it will be able to grow even further and especially in a Linux community where open sources, It's just so vital to develop software and doing collaborate work. That's really where we want to go with this sort of software."

"RWCP High Performance Computing for Structural Biology: Challenging structure biology using large-scale parallel processing" Produced by Prallel Application TRC Laboratory THis video uses various computer graphics to explain structure biology, which is expected to become a fundamental technology of every industry in the 21st century, and explains how paralle computation technology can be used to clarify this field based on the research results of the RWCP Parallel Application TRC Laboratory. The video shows development of various algorithms among at protein structure prediction and the WWW service of the PAPIA syste,.

Prof. George Cybenko Professor of Dartmouth College Chief editor of the IEEE Computational Science & Engineering Magazine

"The applications in DNA analysis and molecular dynamics are going to be extremely important in the 21st century. Kinds of engineering know-hows have been developed for mechanical systems based on mechanical models. However, our future lies in building radical models and doing simulation and design in biotechnology, for drags, for improving agriculture, and for understanding human diseases. These are extremely important application areas. The PAPIA system is used by people around the world by web version, and be extremely popular globally. I think it's outstanding work."

"If prediction of the 3-D structure of proteins becomes possible in principle, fundamental research in protein science will virtually be complete. The next step is to establish how to design desired proteins and the actual process of synthesizing and extracting proteins artificially. If this can be achieved, these techniques will have immeasurable influence."

Prof. Ken Nishikawa National Institute of Genetics

Prof. Constantine Polychronopoulos Professor, University of Illinois Director, Center for Supercomputing Research and Development (CSRD)

"Well there is no question about that is a very important application especially for the biomedical industry, pharmaceutical industry, the research community, and data mining applications. Fast lookup for comparison of traces of molecules verses proteins that maybe store in big databases, it's a very important application. "

I'm very impressed by PAPIA. It's an extremely impressive combination of parallel processing and database research. I'm impressed by the superb technology that be able to apply to really important applications.

Prof. Dennis Gannon Indiana University, Professor

Dr. Toru Yao Adviser, Genome Sciences Center, The Institute of Physical and Chemical Research

"When genome analysis advances further, it can determine all the genes that make up the structure of proteins. For example, there are about 6,000 genes in yeast and about 100,000 in a human being. Since all the structure of proteins is made up and completely determined by genes, the amount of work to be done is tremendous. The task cannot be accomplished by measurement technology alone, and so 3-dimensional structure prediction is necessary. It is a big task to estimate the structure using computers."

"Toward an Advanced Human Interface" Produced by Multi-Modal Functions TRC Laboratory and Information-Base Function TRC Laboratory This video brings together the latest research results of the Multi-Modal Functions TRC Laboratory and Information-Base Function TRC Laboratory, focusing on a notebook multi-modal PC (MMPC) on demonstrations at AAAI-98, an international conference held in Wisconsin from July 26 to 30, 1998. Following our exhibition and demonstration for several kinds of software of an advanced man-machine interface, including a system for recognizing speech and gestures, and speech summarization system, and retrieval system, participants commented as follows. Note: Since we made these on-the-spot interviews without prearrangement, we could not obtain the names and professional background of all respondents, so such information is omitted.

"I'm doing some stuff of gesture recognition in my research. The system demonstrated here seems to work nice in real time. I would like to get a source code and try it for myself."

"It is very impressive that the fact integrated speech and vision images can be recognized in real time on a single computer. It seems like very useful in the future."

"I was attracted by the system dealing with a database of a large amount of images and doing self-organizing and speedy searches. This might be very interesting."

"I saw various demonstrations. This is a good application as it utilizes both speech and visual information."

"I've seen many great demonstrations. But it's quite difficult to put them into practical systems. For example, there may be many problems in recognizing movement regardless of direction if you wave your hand from behind. I think there are a number of issues before a complete system can be built. I hope that you'll overcome such problems and construct an effective system."

"Quite interesting to me because it empirically proves that it can have many functions although it's relatively simple technology."

As stated above, the technological development of RWCP have been winning a worldwide reputation in recent years. For the remaining two years of the project, we will continue to create new technologies that can be used more widely. We will send you a video tape if you are interested in such results as portrayed in the video tapes; send your request to the following. Promotion Department, Real World Computing Partnership 8th floor, Ryukakusan Building, 5-12, Higashi-Kanda 2-chome, Chiyoda-ku, Tokyo 101-0031 Japan Phone: +81-3-5820-8681 Fax: +81-3-5820-8690 E-mail: namikosh@rwcp.or.jp