|
T.Ikedo,Dr. CIS, Hosei University
|
 |
Parims.org has operating through host universities since 1994 for R&D of multimedia engines, focusing on hardware architectures and algorithms especially for graphics processors. The Parims organization is independent from any profit-making corporations and government authorities. Researchers participating in this group include professors, visiting professors, research students, graduate and undergraduate students. New technologies as an intelligent property developed in Parims project exceeded 50 up to the 2010. The primary motivation for our researches is to develop graphics IPs superior to those used in nVidia and ATI's chips. nVidia and ATI have been occupied a top position in the CG market since the 2000, and made great contributions towards CG development. However, the environment that is no one anywhere match for them was less exciting than that of 1980s, when many CG venture companies were in fierce competition for the development of new hardware technology and architectures. As a result, hardware advances in commercial products were not revolutionary, and performance increased simply corresponding to VLSI technologies. Current circumstances, in which almost all CG hardware manufacturers rely upon texture mapping technology developed in 1980s, is the result of such innovation-stifling circumstances. Few research groups in academic field dare to develop hardware architectures. Although there are some international conferences for graphics hardware, many presented papers describe applications using commercial chips (implementation technologies using existent chips), and proposals for truly new architecture are rare, despite that new era of 3D scenography is about to start. Insofar as the industry persists in pushing mapping technology, developers of current chips investigate only how to increase the number of polygons using building-block approaches and how to embed large numbers of texture pattern. Our goal is to restore natural phenomena to the computer world, developing new mobile-able hardware architectures based upon empirical or physical models. To generate a realtime virtual reality scene, fine-grained hardware processing is indispensable. This can be realized only by discovery of hardware algorithms and their implementations, without reliance upon fake or ad-hoc technologies. |
|