| Co-Investigator(Kenkyū-buntansha) |
COLE James B. Univ.of Tsukuba, Inst.of Information Sciences and Electronics, Assistant Professor, 電子・情報工学系, 助教授 (20280901)
ITO Toshiaki Univ.of Tsukuba, Inst.of Information Sciences and Electronics, Assistant Professor, 総合科学学部, 助教授 (60201927)
IKEBE Yasuhiko Univ.of Aizu, School of Computer Science and Engineering, Professor, コンピュータ理工学部, 教授 (10114034)
OYANAGI Yoshio Univ.of Tokushima, Facullty of Integrated Arts and Sciences, Assistant Professor, 理学部, 教授 (60011673)
KAMEDA Hisao Univ.of Tsukuba, Inst.of Information Sciences and Electronics, Professor, 電子・情報工学系, 教授 (10011660)
李 頡 筑波大学, 電子情報工学系, 助教授 (50251046)
|
|---|
| Budget Amount *help |
¥12,900,000 (Direct Cost: ¥12,900,000)
Fiscal Year 2001: ¥5,300,000 (Direct Cost: ¥5,300,000)
Fiscal Year 2000: ¥5,300,000 (Direct Cost: ¥5,300,000)
Fiscal Year 1999: ¥2,300,000 (Direct Cost: ¥2,300,000)
|
|---|
| Research Abstract |
Ever since a central dream of the digital culture has been to create one huge computer. Not a towering superbrain tended by white-coated priests, but a vast constellation of interacting machines - processors, memory modules, disk drives, and a million other devices, all networked into a vast planetary system. A means of thinking, creating, and communicating that is everywhere at once, but nowhere in particular. A computer that is always on. Such a system would continuously spread itself and thicken, expanding by its own internal logic. It would be supremely adaptable, and hard to break. It would have myriad access points, but no CPU, no single point of failure. The global village, to coin a phrase, made real. Take all the intelligent machines in the world - from giant mainframes to the tiniest embedded chip - andhook them together in a single intelligent network. A system open to novelty, new members, and features. A system that can tolerate what engineers ruefully call faults. Asystem
… More
with no limits on how large it can get, nor how small its smallest part can be. The key pieces for such a system - millions and billions of microprocessors - are already here, or coming. So, too, are the riotously expanding networks. Indeed, to start building that one great computer, only a single essential ingredient is missing: an architecture, a universal language, a set of superprotocols, something - and very possibly today's lexicon can't name it - to hold it all together and let the magic work. A constitution, if you like, a digital equivalent of the genetic code that all living things share. A three-dimensional full electromagnetic particle-in-cell (PIC) code, TRISTAN (Tridimensional Stanford) code, has been parallelized using High Performance Fortran (HPF) as a RPM (Real Parallel Machine) and tested on a networked commodity PC microcomputers. In the simulation, the simulation domains are decomposed in one-dimension, and both the particle and field data located in each domain that we call the sub-domain are distributed on each processors. Both the particle and field data on a sub-domain is needed by the neighbor sub-domains and thus communications between the sub-domains are inevitable. Our simulation results using HPF exhibits the promising applicability of the HPF communications to a large scale scientific computing such as 3D particle simulations. Less
|
