Biosupercomputing symposium 2008 in Tokyo

The Biosupercomputing Symposium 2008 was held on 12/25 - 26 in Tokyo by RIKEN sponsorship. I found that it was the one of the most interesting HPC symposiums in Japan recently. The MY PLAZA hall in Marunouchi, Tokyo was almost full by many attendees in spite of the last business day of 2008.

This symposium was more than just for presentations about progress in RIKEN's Next-Generation Integrated Life Simulation Project targeted into the next generation supercomputer of Japan. It included a keynote, a poster session by younger researchers, and invited talks.

I am most interested in the presentation by the brain science team (a cerebral nerve system research and development team) that was added newly from this October in the project. A team leader, professor Shin Ishii, Kyoto University, introduced that their objective are to develop models and simulation software that could simulate the relationship between input and output in whole brain level at first in their long-term goals of the project, i.e., Grand Challenge.

Direct contribution to the medical treatment is out of scope because of a short project period according to Prof. Ishii's comment to an attendee's question.

There were 76 teams participating in the poster session led by young researchers studying in Japan. This may indicate that the RIKEN's Next-Generation Integrated Life Simulation Project well penetrates universities and institutes. Naoto Yamamura, RIKEN won the best poster award by "Development of the skeletal muscle simulator", and five teams won poster award.

There was a call for establishing Biosupercomputing Research Community (bscrc) by Prof. Haruki Nakamura, Osaka university. He encouraged attendees to join the bscrc as a promoter. We can apply it in the home page of Biosupercomputing Research Community.


By the way, the Next-Generation Integrated Life Simulation Project is a comprehensive modeling/simulation project for organism that anyone could not challenge before. Therefore, it becomes important to keep it visible about the integrated view with a variety of R&D components in order to enable effective interactions between project members and researchers in academic medical research centers/pharmaceutical industry.

It is often seen that the successful Japanese life sciences projects well involve both researchers in science/engineering and medical doctors in effective manner. I observe that it is similar in Europe and U.S.A, such as in Blue Brain Project. We would expect more such opportunities in Japan.

The SC08 report by RIKEN's SC08 Reporters

･ The report of SC08 is now in publish on the web site of RIKEN's next-generation supercomputer R&D Center.

Three young reporters are those who won the highest award and two outstanding performance awards in the poster session of next generation supercomputing symposium 2008 in Tokyo:

- Takayuki Saito (National Astronomical Observatory, Japan)
- Katsumi Hagita (National Defense Academy in Japan)
- Koji Makino(Research Organization for Information Science &
Technology, Japan)

･ The approximately 70 page-report looks a lot of efforts, and it reflects youthful zest of each of three young reporters.

･ It must be a very good program to stimulate younger HPC people. I hope that I can enjoy the SC09 report next year very much.

The iPAB open seminar "Trend and future of HPC"

The open seminar of the initiative Parallel Bioinformatics(iPAB) was held on December 5, featured "Trend and future of HPC". This open seminar was focused on high performance computing (HPC) indispensable to bioinformatics.

Besides iPAB, Chem-Bio Informatics Society (commonly called CBI Gakkai) with a long history, and Japanese society for bioinformatics (JSBI) that held the annual meeting in Senri on 15th - 16th of December are known as an academic active group in Japan which intends to involve HPC for the life science.

They have their own characteristics, and in my impression, iPAB seems to have the keen interest in an immediate application of HPC, JSBI is active in education and basic research of bioinformatics, and CBI Gakkai works in broad area relative to the drug discovery.

Well, three following HPC lectures were literally done with the theme in "Trend and future of HPC" open seminar in the O-okayama campus of Tokyo Institute of Technology by iPAB. Because of that, it was few attendees from the life science area, and many of participants are related to HPC.

Lecture 1. "SC08 report" by Prof. Yoshio Oyanagi, Kogakuin University

･ Prof Oyanagi's SC08 annual report are revised on 12/7.

Lecture 2. "Trend and future of HPC" by Prof. Taisuke Boku, Tsukuba University

･ It was mainly a lecture on technology trends of the cluster system. It was a good and comprehensible lecture taking T2K-Tsukuba large-scale, general-purpose cluster for example (T2K means Tsukuba, Tokyo and Kyoto).

･ It is completely agreed, in particular, for the point that the gap between the floating point arithmetic performance and the memory bandwidth is extending more and more because the processor with an extremely high floating point arithmetic performance has come to be used with a commodity HPC cluster system and it becomes a serious issue. For observation that many-core chip could not significantly contribute to HPC as a result, Prof. Taisuke Boku suggested that surplus cores might have to be used for the purposes other than the floating point calculation.

･ Moreover, about accelerator of much interest in SC08, he said that because accelerator is comparatively easy to implement on the cluster system , the aggressive use can be just conceivable. In that case, the data transfer bandwidth to accelerator must be the severest issue. Also the issue of the reliability of GPGPU (e.g., There is no ECC in a built-in memory of the GPU board) was discussed.

Lecture 3. "The latest TSUBAME system", Prof. Satoshi Matsuoka, Tokyo Institute of Technology

･ It was a lecture about the TSUBAME 1.2 system of Tokyo Institute of Technology. TSUBAME is known well as a system that has positively taken accelerator at first and now 648 ClearSpeed accelerators are being attached with it according to Prof. Matsuoka. In addition, the performance was improved by enhanced 318 nodes added Tesla S1070 accelerator. Then TSUBAME won the second place in Japan succeeded to the T2K of University of Tokyo in the Top500 list of November, 2008.

･ Thus the node of TSUBAME 1.2 has three types: (A) Opteron + ClearSpeed + Tesla (318 nodes), (B) Opteron + ClearSpeed (330 nodes), and (C) Xeon (90 nodes). It is a really heterogeneous commodity cluster system. It may be beyond expectation in general that such a heterogeneous system realizes better LINPACK performance in short period after the version up, and typical cluster users could never do something like that.

･ According to Prof. Matsuoka, GPU resembles a classical vector computer because of the feature of the high peak performance and a high memory and the bandwidth (in GPU). However he also pointed out that GPU also has some technical limitations and further study is required as shown in the history of PC cluster becoming main stream of a general-purpose parallel computer. Such situation imagines me the enough possibility that some Japanese computer manufacturer might develop their original accelerator in the future.

･ TSUBAME 2.0 is supposed to be run in the spring of 2010. Do not miss it in one and a half years.


Although I did not know details of TSUBAME, the previous talk by Prof. Bok helped me to spend interesting time to understand TSUBAME much well. Well, except for those having high level skills and experiences enough to achieve implementation like Tokyo Institute of Technology, it looks that the entire cost will never become cheap of the expectation from the word commodity cluster system for common users excluding particular universities. In addition, since application porting/tuning efforts are mandatory for such a peta-scale cluster systems to run applications with reasonably high performance, the word challenge is more suitable than exploiting. Grand Challenge is just exact word.

Another in SC08

･ Now I like to add another with some photos that I couldn't write because of not enough time to write during SC08.

BG/P Board

･ First, a board being put casually in the booth of IBM is a board of the Blue Gene/P. There are 32 computational node(processor)s and 2 I/O node(processor)s on the board. A power supply places around center line. A four core-SMP is inside a processor chip in the Blue Gene/P. The performances per board is 435 GFLOPS in double/single precision. Its reliability (MTBF) is 10 to 100 times higher than commodity-based cluster systems. The Blue Gene satisfies exact MPP specifications after all.

iDataPlex

･ In contrast, IBM iDataPlex is the latest commodity-based cluster by IBM. It looks two racks because the iDataPlex rack is designed using depth of a common rack for width and width for depth approximately in order to improve heat removal capability. It is a state-of-the-art cluster of high density and the high cooling efficiency.


NCAR

・ Next is a panel of the Bluefire system that was exhibited in the booth of NCAR, a leading weather research institute. This is water-cooled cluster system using IBM Power 575 with 4.7GHz POWER6 processor. Many of weather research institutes are using IBM Power 575 cluster system in the world.






Rackable

･ This is a system of the container type by the Rackable that facilitates cluster system and air conditioning in the trailer. It can move and start a data center anywhere. The wheels are seen under the trailer. The wind of air-conditioning blew to my face when entering inside. The passage is arranged in center and servers and equipment are rack-mounted on the wall side. Since such a data center can start up in a short term in a disaster or a big event, it might be very convenient for the purpose.




NVIDIA

･ The following is a NVIDIA booth that crowded in the first day. The personal computer that was enhanced with Tesla GPGPU was exhibited.The signboard explains it as almost one million yen for 960 cores and 3.732TFLOPS.
It is certainly surprising but the performance figure is only for single precision operation. Many researchers variously pointed out when Cell/B.E processor was announced, if double precision arithmetic is not very fast, It is not a practical solution in the large-scale HPC computational use. Because a huge number of arithmetic operations gradually degrade accuracy of calculation due to rounding error and so on, and a single precision cannot remain reasonable amount of digits in typical peta-scale calculation at least.

SONY

･ SONY showed a new server including their Cell/B.E processor and NVIDIA RSX (GPU) in the booth of Argo Graphics. They demonstrated a real time visualization with AVS parallel version of the KGT Co. while simulation runs on the cluster consists of four servers as shown in the photo. It is interesting that such a HPC system is provided comparatively affordable price (but still order of million yen).Even in the Cray booth, a super computer on the desk side was exhibited.





We can imagine that these affordable HPC system drags younger students to HPC in future, and they can actually get HPC skill with such a system, can't we?