论文信息 - PAD cluster: an open, modular and low cost high performance computing system

PAD cluster: an open, modular and low cost high performance computing system

 This work presents the PAD Cluster, the result of PAD Elebra Project. It was built on COTS (commodity off-the shelf) components. It is shown the hardware and software architecture of the cluster. Some performance results are presented in the implementation of an user level communication library. Cluster management tools for a single point of control are also presented. Keywords High performance computing, Cluster of workstations, parallel processing

Sergio Takeo Kofuji | Marcio Lobo Netto | Volnys Borges Bernal | Guilherme Matos Sipahi | A. G. Anderson

[1] Henry G. Dietz,et al. A fine-grain parallel architecture based on barrier synchronization , 1996, Proceedings of the 1996 ICPP Workshop on Challenges for Parallel Processing.

[2] Thomas L. Sterling,et al. BEOWULF: A Parallel Workstation for Scientific Computation , 1995, ICPP.

[3] Sergio Takeo Kofuji,et al. The barrier synchronization impact on the MPI-programs performance using a cluster of workstations , 1997, Proceedings of the 1997 International Symposium on Parallel Architectures, Algorithms and Networks (I-SPAN'97).

[4] Walt Wimer. Clarifications and Extensions for the Bootstrap Protocol , 1993, RFC.

[5] Peter S. Pacheco. Parallel programming with MPI , 1996 .

[6] Charles L. Seitz,et al. Myrinet: A Gigabit-per-Second Local Area Network , 1995, IEEE Micro.

[7] Sergio Takeo Kofuji,et al. Avaliação do U-Net em clusters com rede Myrinet , 1998 .

[8] Márcio Lobo Netto. Development of high performance parallel graphical applications with efficient parallel processing and adaptive techniques , 1998 .

[9] Thorsten von Eicken,et al. ATM and fast Ethernet network interfaces for user-level communication , 1997, Proceedings Third International Symposium on High-Performance Computer Architecture.