| URGE to Compute Wiki |
|
URGE2012 »
Tools And MethodsMondays 5:30-7:30pm Day 1Monday, Jan 23, 2012: Outline Day 2Monday, Jan 30, 2012: Outline Day 3Monday, Feb 6, 2012: Outline Day 4Monday, Feb 13, 2012: Shell scripting (Dave Yearke, Director, Science and Engineering Node Services) Day 5Monday, Feb 20, 2012: (Presenter David Yearke, Director, Science and Engineering Node Services) Day 6Monday, Feb 27, 2012: Python Scripting (John Ringland) Day 7Monday, March 5, 2012 Day 8Monday March 26, 2012 Day 9Monday April 2, 2012 Day 10Monday April 9, 2012: T&M double-header
Day 11Monday April 16, 2012 Assignment Each student should run the mpi (./build4mpi) and OpenMP (./build4omp) versions of the Julia Set application. Do this by requesting an interactive job that has 12 processors per node (ppn=12 in the qsub command). Keep track of the elapsed time reported by each version and compare with the one reported when you used the cpu (i.e. serial) version. Hint: Adding the -q debug option to your qsub command will request access to the debug queue and reduce the time you have to wait for your interactive job to be scheduled. Calculate the speedup for each version (speedup=elapsed_time_serial/elapsed_time_parallel) and e-mail your results to lsmatott@buffalo.edu. In your e-mail you should also include a discussion of the speedup - is it about what you would expect, given that the parellel versions use 12 processors vs. just one for the serial version? Why or why not? Day 12Monday April 23, 2012: Bibliographic management Day 13Monday April 30, 2012 |
General URGE 2013 URGE 2012
URGE 2011
URGE 2010
|
| Page last modified on May 01, 2012, at 12:02 PM - Powered by PmWiki |