1 / 15

Oregon Chub Beowulf Cluster

Oregon Chub Beowulf Cluster. Authors A.J. Supinski Billy Sword. Advisor Dr. Rylander, Dr. Lillevik Industry Representative Mr. Noah Van Dresser Intel Corp. Agenda. Introduction A.J. Supinski Background A.J. Supinski Methods A.J. Supinski Results Billy Sword

garron
Download Presentation

Oregon Chub Beowulf Cluster

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Oregon Chub Beowulf Cluster Authors A.J. Supinski Billy Sword • Advisor • Dr. Rylander, Dr. Lillevik • Industry Representative • Mr. Noah Van Dresser • Intel Corp. University of Portland School of Engineering

  2. Agenda • Introduction A.J. Supinski • Background A.J. Supinski • Methods A.J. Supinski • Results Billy Sword • Conclusions Billy Sword • Demonstration Billy Sword University of Portland School of Engineering

  3. Introduction Thanks to Dr. Rylander and Noah Van Dresser for all of their help. Overview. Allowing non-homogenous clusters via Genetic Algorithms (GAs). Cost efficiency increase for those who use non-homogenous clusters. University of Portland School of Engineering

  4. Background Beowulf Clusters are sets of computers which have software that makes them act like one supercomputer. Process Scheduling is an NP-Complete problem. GAs have been shown to produce good (non-optimum) solutions to NP-Complete problems in polynomial time. Our project is to check for speedup from existing cluster process scheduling to the GA scheduling. University of Portland School of Engineering

  5. Methods The Plan: Build a non-homogenous Beowulf cluster, benchmark it and then modify it to use GA scheduling and perform the benchmarks again. Ensure no change in environment by using the same hardware and almost entirely the same software for both tests. Results will be restricted by limitation of 4 PCs. University of Portland School of Engineering

  6. Methods The Action: Build Hardware 4 PC cluster. RedHat Linux 9.0 OSCAR (Open Source Cluster Application Resources) . PXE Boot Kernel issues. MPICH (A Portable Implementation of Message Passing Interface) University of Portland School of Engineering

  7. Results Working 4 PC Beowulf Cluster running MPICH. Working GA prototype but not in MPICH. MPICH software architecture challenges. University of Portland School of Engineering

  8. Life University of Portland School of Engineering

  9. University of Portland School of Engineering

  10. Mastermind University of Portland School of Engineering

  11. University of Portland School of Engineering

  12. Fractal Image University of Portland School of Engineering

  13. Conclusions The use of GA processor scheduling remains a viable idea MPICH modification would require a lot of work. This might be possible as a future Senior Design project starting from our point of ending. We have a Working Beowulf Cluster and good instructions for future attempts. University of Portland School of Engineering

  14. Demonstration University of Portland School of Engineering

  15. Questions? University of Portland School of Engineering

More Related