From kbuyukat@mail.coin.missouri.edu Thu Jun 13 23:01:34 1996 Date: Thu, 13 Jun 1996 20:13:26 -0500 (CDT) From: Kaya Buyukataman To: ** ITUMD ** ISTANBUL TEKNIK UNIVERSITESI MEZUNLARI DERNEGI ULUSLARARASI KURULUSU Subject: Efficient Parallel Algorithms for Reconfigurable Mesh Architectures (fwd) _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ _/ To: ITU Alumni Scattered All Around The World _/ _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ _/..ITU-MD................ ISTANBUL TEKNIK UNIVERSITESI _/ _/........ /\ /\ ..... _/ _/..... / | / | .... Mezunlari Dernegi-Alumni Association _/ _/.. / // | ... _/ _/. ^'^'/ / / .. _/ _/. /()|'`--/ _ / .. Adr: 235 E River Dr. Suite 1502 _/ _/. \_'/' '` \_' .. E Hartford, CT, 06108, USA _/ _/.. '//`--' '\ .. Phone-Fax : (203) 282-0251 _/ _/... || //--/ ' \ .E-mail DagItIm _/ _/ _/..... \\ ||\_/\\ ... E-mail Int: _/ _/....... || \_/ .... E-mail TR: _/ _/.......1989 \\ \ .... http://www.ece.orst.edu/~ituweb _/ _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ _/ ** COMPONENT ALUMNI ASSOCIATION LOCATIONS ** _/ _/Almanya-Ingiltere-Turkiye(2)-Japonya-Kanada(3)-Benelux-Avusturya-Isvec_/ _/ Avustralya-Suudi Arabistan-Amerika(16)-Norvec-Hongkong-Fransa-Kibris _/ _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ From: Ilker Ersoy IEEE - ITU Ogrenci Kolu Efficient Parallel Algorithms for Reconfigurable Mesh Architectures Doc. Dr. Fikret ERCAL Department of Computer Science University of Missouri-Rolla Rolla, MO 65401 Tarih : 26 Haziran 1996 Saat : 14:00 Yer : Elektrik Elektronik Fakultesi Video Odasi (D1302) Abstract: Most of the currently available parallel computers are based on a collection of processors connected via a fast interconnection network with a fixed topology. An attractive interconnection scheme is the 2D mesh-connected architecture because of its simplicity and scalability. On the other hand, the communication diameter for an NxN mesh is N and therefore, it becomes a bottleneck in solving problems efficiently. It is shown thatreconfigurable mesh (RMESH) architectures offer the needed efficiency and flexibility in interprocessor communications by allowing unit-time broadcasts broadcasts on a network whose topology can change dynamicall as required by the algorithm. The RMESH architecture and some efficient parallel algorithms developed for it, such as maze routing and approximate string matching, will be presented during this talk. ========================================================================