Details

Written by Douglas Eadline

Published: 16 May 2007

Hits: 8405

Version 4.2 of the GCC tool-set was just released. This release includes OpenMP which is an Application Program Interface (API) for shared-memory parallel programming in C/C++ and Fortran. OpenMP is jointly defined by a group of major computer hardware and software vendors and is a portable, scalable model that gives shared-memory parallel programmers a simple and flexible interface for developing parallel applications. While OpenMP has been around for a while, the growth of multi-core processors has made tools like OpenMP valuable to software developers. OpenMP provides a series of directives that can be added as comments to existing programs (i.e. it is easier than threaded programming). The inclusion of OpenMP into the GNU tool-set should help with multi-core programming, but it is not an effective way to create codes that use a distributed memory model (like clusters). MPI is still the best method for this kind of programming. Hybrid programming, OpenMP used on multi-core nodes and MPI used to communicate between nodes is certainly possible. Any takers?

Details

Written by Douglas Eadline

Published: 30 April 2007

Hits: 3089

intel -see links on Slashdot http://www.pcper.com/article.php?aid=382 http://arstechnica.com/news.ars/post/20070328-intel-aims-nehalem-at-amds-fusion-integrated-graphics-on-die-memory-controller-smt.html http://techreport.com/reviews/2007q1/cpus/index.x?pg=1 http://www.zdnetasia.com/news/hardware/0,39042972,62021390,00.htm http://techreport.com/onearticle.x/12127 http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2955 AMD - http://www.tgdaily.com/content/view/31729/135/ http://www.informationweek.com/news/showArticle.jhtml?articleID=199501467&pgno=1&queryText= http://www.informationweek.com/news/showArticle.jhtml?articleID=199501467 Comparison: http://techreport.com/reviews/2007q2/pricevperf/index.x?pg=1 Cell - sedongara paper http://www.netlib.org/utk/people/JackDongarra/PAPERS/scop3.pd http://charm.cs.uiuc.edu/papers/CellPMUP06.psf

Details

Written by Douglas Eadline

Published: 09 April 2007

Hits: 7762

Want to become a cluster ninja and move one step closer to that world domination goal? If you answered, yes, your opportunity awaits this Spring and Summer. One might consider this a shameless plug, but in the interest of world domination (yours not mine) I invite you to take a look at the upcoming raft of advanced cluster short courses. These courses are brought to you by the ARC at Georgetown University. More information can be found at the ARC HPC Training Page. Here is a list of the scheduled courses:

• April 17-20 Introduction Beowulf Design, Planning, Building and Administration
• June 4-5 Advanced Sun Grid Engine
• June 6-8 Intermediate Beowulf Administration and Optimization
• June 11-13 Advanced Condor Job Scheduler Configuration and Administration
• July 16-20 Introduction to Parallel and Distributed Programming

Class size is limited, so if you have interest, sign up early. I'm slated to teach the Intermediate Administration and Optimization course. I grade on a curve.

Details

Written by Dan Stanzione

Published: 01 April 2007

Hits: 9040

From the Ethernet rules the day, department

We are pleased to announce the new software package developed by a team of Cluster Monkey researchers including Robert Brown, Jeff Layton, and Douglas Eadine. The project was sponsored by the Beowulf Legacy Enhanced Ethernet Project (BLEEP). The results of the these efforts is a new software package called Geyser™ that will be available for Linux later this month. By substituting the Geyser™ drivers in the Linux Kernel, a standard Ethernet card can transmit at more than double the standard Gigabit Ethernet rate. Moreover the single bit latencies are less then one microsecond. Please read the theoretical background to fully understand this report.