| Presentation | 2019-07-26 Inter-Node Direct Memory Access Performance with Distributed Switch for Full-Mesh Connection Between Accelerators with Optical Hub Kenji Mizutani, Hiroshi Yamaguchi, Yutaka Urino, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In order to improve communication performance for paralle processing, we have proposed an optical hub that connects nodes by full-mesh network . In this paper, we introduce the distributed switch with a simple circuit switching mechanism to construct an optical hub. By using this distributed switch, in the communication process of the sending node send, we achieved the low latency result (0.34 usec) in one point-to-point memory-to-memory communication. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | parallel distributed processing / communication / full mesh connection / accelerator / tight coupling / point to point communication / delay |
| Paper # | CPSY2019-32,DC2019-32 |
| Date of Issue | 2019-07-17 (CPSY, DC) |
| Conference Information | |
| Committee | CPSY / DC / IPSJ-ARC |
|---|---|
| Conference Date | 2019/7/24(3days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Kitami Civic Hall |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Parallel, Distributed and Cooperative Processing Systems and Dependable Computing |
| Chair | Hidetsugu Irie(Univ. of Tokyo) / Satoshi Fukumoto(Tokyo Metropolitan Univ.) / Hiroshi Inoue(Kyushu Univ.) |
| Vice Chair | Michihiro Koibuchi(NII) / Kota Nakajima(Fujitsu Lab.) / Hiroshi Takahashi(Ehime Univ.) |
| Secretary | Michihiro Koibuchi(Nagoya Inst. of Tech.) / Kota Nakajima(Hokkaido Univ.) / Hiroshi Takahashi(Nihon Univ.) / (Chiba Univ.) |
| Assistant | Eiji Arima(Univ. of Tokyo) / Shugo Ogawa(Hitachi) |
| Paper Information | |
| Registration To | Technical Committee on Computer Systems / Technical Committee on Dependable Computing / Special Interest Group on System Architecture |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Inter-Node Direct Memory Access Performance with Distributed Switch for Full-Mesh Connection Between Accelerators with Optical Hub |
| Sub Title (in English) | |
| Keyword(1) | parallel distributed processing |
| Keyword(2) | communication |
| Keyword(3) | full mesh connection |
| Keyword(4) | accelerator |
| Keyword(5) | tight coupling |
| Keyword(6) | point to point communication |
| Keyword(7) | delay |
| 1st Author's Name | Kenji Mizutani |
| 1st Author's Affiliation | Photonics Electronics Technology Research Association(PETRA) |
| 2nd Author's Name | Hiroshi Yamaguchi |
| 2nd Author's Affiliation | Photonics Electronics Technology Research Association(PETRA) |
| 3rd Author's Name | Yutaka Urino |
| 3rd Author's Affiliation | Photonics Electronics Technology Research Association(PETRA) |
| Date | 2019-07-26 |
| Paper # | CPSY2019-32,DC2019-32 |
| Volume (vol) | vol.119 |
| Number (no) | CPSY-147,DC-148 |
| Page | pp.pp.171-176(CPSY), pp.171-176(DC), |
| #Pages | 6 |
| Date of Issue | 2019-07-17 (CPSY, DC) |