| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: The Clean Energy Project - Phase 2 Forum
Thread: Research Log: Updates from the Harvard Team |
| No member browsing this thread |
|
Thread Status: Active Thread Type: Sticky Thread Total posts in this thread: 1840
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
BTW: it is worth noting that Roel Sanchez who was with the CEP team in the beginning won the same ACS CINF award a few years ago as well. Nice track record.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Our friends from IBM WCG already posted this elsewhere, but we thought it would make sense to post it here on our log as well: We currently have rather large molecules on the grid and our error rate started to increase notably. The reason for this is that there is an accumulation of rounding errors for these big compounds, and the results ended up above the validation threshold – although they were perfectly fine. We have now cranked up the threshold a little to clear these cases. Sorry for any inconvenience this may have caused.
Best wishes Your Harvard CEP team |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The Harvard CEP team wishes you all a great weekend. Enjoy the nice weather as long as it lasts – and let your computers crunch away on the CEP in the meantime :).
----------------------------------------[Edit 1 times, last edit by Former Member at Sep 14, 2013 5:36:06 PM] |
||
|
|
Rickjb
Veteran Cruncher Australia Joined: Sep 17, 2006 Post Count: 666 Status: Offline Project Badges: 
|
Welcome home, CE, and congratulations on your successful presentations at the conference. I'd been wondering about your period of silence here ...
"We currently have rather large molecules on the grid and our error rate started to increase notably. The reason for this is that there is an accumulation of rounding errors for these big compounds, and the results ended up above the validation threshold – although they were perfectly fine. We have now cranked up the threshold a little to clear these cases. Sorry for any inconvenience this may have caused." I did notice that the number of CEP2 error results from my machines had decreased - read "stopped" - but I found a couple where I've been sent a repair WU when 2 of the wingmen have already returned potentially valid results, and was wondering why I was sent the extra copy. But instead of having status of "Pending validation", these wingmens' WUs are "Pending verification". Does this indicate that the difference in their results is over a threshold but not unreasonable, so another result is sought as a check? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Martin is testing out the new Odyssey2 cluster of Harvard FAS Research Computing these days. The FAS RC team has done an outstanding job putting this new high-performance hardware together and it is a great complement to our work on the World Community Grid. People have been asking why we use both clusters and the grid. The answer is that these are very different types of resources. The grid is incredible when it comes to volume computing, i.e., the large scale generation of results, but it has its limitations in terms of performance. A cluster on the other hand is specifically designed for hard tasks and results you need on short notice, but one could basically never get computing time at the scale provided by the WCG. A last point in this consideration is that the grid - as a virtual supercomputer - is a free resource, i.e., it didn’t need to be built or bought for our project but utilizes otherwise idle and wasted CPU cycles. That is one of the coolest aspects of it.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi Rickjb,
yes, it's great to be back to normal :). About the "verification" vs "validation" - we are actually not quite sure about the difference in terminology. This is on the WCG end. Maybe they can chime in on this? Best wishes Your Harvard CEP team |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7696 Status: Recently Active Project Badges:
|
Think of the verification check as a machine integrity check, more or less.
----------------------------------------Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Rickjb
Veteran Cruncher Australia Joined: Sep 17, 2006 Post Count: 666 Status: Offline Project Badges:
|
I think the verification check is the equivalent of calling for the opinion of the 3rd umpire in sport (eg cricket) ...
For one of the WUs in question, a 3rd result has been returned, and this plus 1 of the pair that were Pending verification have been marked Valid, while the other one that was pending verification has been marked Invalid. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
We recently had a big push to update and revamp many of the codes that form the automated, high-throughput infrastructure of the Clean Energy Project.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
We found a new research partner at Oxford University in England. So, by now we have accumulated a quite impressive set of collaborators for the Clean Energy Project.
|
||
|
|
|