Page 1 of 1

Massive computing project to develop magnetic fusion

Posted: Sun Jan 18, 2009 3:42 am
by PolyGirl
An article over at Physorg.com Titled "Scientist receives massive computing project award to develop magnetic fusion energy"
Choong-Seock Chang, a research professor at New York University's Courant Institute of Mathematical Sciences, has received a Department of Energy (DOE) award to carry out ultra large-scale computation using the Cray XT supercomputer at the department's Oak Ridge National Laboratory in Tennessee. The awarded 20 million hours of computing time—roughly equivalent to running a single-processor desktop computer for more than 2,280 years—is among the largest awards given to a single project. The computation will be using more than 100,000 processors at a time.
However this project is probably more weighted for ITER than the Polywell Fusion. Maybe Dr Nebel can find out more about the project and see if the Polywell Fusion can benefit from this research.

Regards
Polygirl

Posted: Sun Jan 18, 2009 3:57 am
by JohnP
The awarded 20 million hours of computing time—roughly equivalent to running a single-processor desktop computer for more than 2,280 years
Errr.... 20 million hours = 2280 years. Was he awarded 20M cpu-hours for the whole system, or 20M hours divided by 100,000 CPU's? Or ???

Posted: Sun Jan 18, 2009 5:46 am
by PolyGirl
Approximately 200 hours for the Cray XT

(24*365*2283)/100,000 = 200hrs or 9 days Approximately

Regards
Polygirl

The true Report

Posted: Sun Jan 18, 2009 10:27 am
by PolyGirl
I should have known better than to jump at news articles like this “Mea culpa”.

Anyway after some digging I found some document / articles in relation to this news article and are listed in a chronicle fashion:
  • This document FSPWorkshipReport relates to one of many about Fusion Simulation Project (FSP). If you read section "1.4 Deliverables", an interesting deliverable was “A range of models that include fundamental computations”, two paragraphs down it mentions, “at the end of 10 years” and “at the end of 15 years” of added capabilities.

    This announcement from the DOE states “The purpose of the present Program Announcement is to competitively select a nationally coordinated interdisciplinary team consisting of fusion scientists and technologists, applied mathematicians, computer scientists, and computational scientists to carry out a one- to two-year detailed planning study for the FSP.”

    This announcement just advises of a different closing date.

    This article tells us who won the contract for the first year.
So other than Choong-Seock Chang winning the contract (whose name also appears in the FSPWorkshopReport.pdf) for a feasibility study, some meaningless numbers being added and some computational coding for the ITER this was the sum total of the “scientific report”.

Regards
Polygirl