| Project/Area Number |
62550254
|
|---|
| Research Category |
Grant-in-Aid for General Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
計算機工学
|
|---|
| Research Institution | Yamagata University |
|---|
Principal Investigator |
TANNO Kuninobu YAMAGATA UNIV.ASSOC.PROF., 工学部, 助教授 (40007018)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
YAMAMOTO Hiroaki SHIHYUU UNIV.ASSOC.PROF., 工学部, 助教授 (10182643)
|
|---|
| Project Period (FY) |
1987 – 1989
|
| Project Status |
Completed (Fiscal Year 1989)
|
| Budget Amount *help |
¥1,900,000 (Direct Cost: ¥1,900,000)
Fiscal Year 1989: ¥200,000 (Direct Cost: ¥200,000)
Fiscal Year 1988: ¥400,000 (Direct Cost: ¥400,000)
Fiscal Year 1987: ¥1,300,000 (Direct Cost: ¥1,300,000)
|
|---|
| Keywords | Parallel computer / Computer networks / MIMD-type computer / SIMD-type computer / Data flow computer / Communication protocol / High-speed FFT / ト-クンリング / 計算機結合網 / データフロー計算機 / 並列ソート / 二重二進木の埋め込み / バイトニックソート / 奇偶変換ソート / 人工知能計算機 / 経路選択 / MIMDマシン / 環状ネットワーク / トークンリング方式 / 並列処理 |
|---|
| Research Abstract |
In this research, we have made an effort to develop a general purpose parallel computer combining and harmonizing processing and communications. our results are as follows;
(1)Development of the data flow computer "Genryuu"using a ring network: The data flow computer called "Genryuu"is composed of a management processor (HOST)and two processing elements connected together on a ring network. The computer provides a original data flow language SAL(Single Assignment Language) to write programs for the machine. It is just a prototype, but we proud that we completed the data flow computer during the three years research.
(2)Analysis of a eight-neighbor mesh-connected processor array: We constructed the eight-neighbor mesh-connected processor array composed of many ring-type processing elements which consist of a processing unit and eight switches connected together on a ring. Then the processor array has eight links, that is, degree 8. As an analytical result, the array with wraparounds is superior to the binary n-cube if the number of processors is under 256.
(3)Development of two fast parallel sorting algorithms and a fast radix 4 parallel Fourier transform algorithm: Two fast parallel sorting algorithms adopting the bitoni sort and the odd-even transposition sort algorithms to the eight neighbor processor array described above. We also developed the fast parallel FFT on the eight neighbor processor array using the radix 4 FFT algorithm. Then we made the computation time two times faster than that of the conventional algorithms both in the sorting and the FFT.
(4)Development of a self-token ring access protocol: We developed a new ring access control scheme to support high-speed communications between the ring processing elements on the eight neighbor mesh-connected processor array. In this protocol, each processor on the ring has a private token, called self-token. We had a result that the protocol is suitable for 1 Gbps high speed communications.
|
