| Project/Area Number |
10044117
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B).
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
計算機科学
|
|---|
| Research Institution | TOHOKU UNIVERSITY |
|---|
Principal Investigator |
NAKAMURA Tadao Graduate School of Information Science, Tohoku University, Professor, 大学院・情報科学研究科, 教授 (80005454)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
FUKASE Masa-aki Faculty of Science and Technology, Hirosaki University, Professor, 理学部, 教授 (10125643)
KOYANAGI Mitsumasa School of Engineering, Tohoku University, Professor, 大学院・工学研究科, 教授 (60205531)
HASEGAWA Katsuo Graduate School of Mathematics, Nagoya University, Professor, 多元数理科学研究科, 教授 (70004463)
KOBAYASHI Hiroaki Graduate School of Information Science, Tohoku University, Associate Professor, 大学院・情報科学研究科, 助教授 (40205480)
HAGIWARA Masafumi Faculty of Science and Technology, Keio University, Associate Professor, 理工学部, 助教授 (80198655)
|
|---|
| Project Period (FY) |
1998 – 1999
|
| Keywords | Computer Architecture / Instruction-Level Parallel Processing / Computer Graphics / Neural Networks / Brain Models / Memory / Volume Rendering |
|---|
| Research Abstract |
In addition to high speed data processing by supercomputing, introducing of the functions of the human brain has contributed to build flexible computers. Structural hints from the human brain allowed us to think about more flexible functions that lead to various data/information processing. In this research we have improved the architecture of brain-structured computers, designed an MISD processor, and implemented an example of the design. We call this architecture the SHIFT MACHINE architecture which can be regarded as an MISD computer. This work is based upon several research results consisting of the analysis of VLIW architectures with a cache analysis and a speculation method. The simulator of the SHIFT MACHINE is available to show its behavior visually. Related to the SHIFT MACHINE architecture, we have also developed a reconfigurable synchronous dataflow computer called SOUND. This was implemented in a chip processor fashion and evaluated. As a result, the left-brain function has
… More
been implemented and the speed has been accepted to be a suitable one under low power condition. On the other hand, the right brain has been implemented in the research of computer graphics and volume rendering. These two subjects has been realized speedy and flexibly by improving the algorithms. In computer graphics of our system, the algorithm is evaluated on a commercially available parallel machine and we got the fast rendering. This fact is proved in volume rendering. Developing reasonable algorithms for computer graphics and volume rendering, we can reach some potential to show fast rendering that is suited for real time processing and rendering. Further neural network research has been developed to discover the right brain function. The results are worth comparing artificial mechanism with the human brain structure.
From these two results in left-brain and right brain research, we have discussed the integration of these two brain functions in terms of the mutual behavior of the functions of the left and right brains. To have these two get together, we concluded that the integration is based on the processing speed on computers engaged in the left and right brains. To increase their processing speed, we have developed the architecture with software including special speculation. Also a cache mechanism has been developed to have high speed processing. Less
|
