| 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 |
Thanks to Suleyman, jabba4 with 135TB of data capacity is now commissioned and we have a lot of extra space for CEP again. It actually was about time to get the new array up and running since we were already using overflow loaner space provided by our friends at Harvard Research Computing. (Thanks a lot to FAS-RC for helping us out on this BTW.)
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I really appreciate the milestone updates (primary results, run-time years) and was wondering if you have an overall count of # of molecules inspected?
I remember reading in Wired that one goal was to inspect 3.5 million molecules, where are we at today and will that number increase for 2013 and 2014? Thanks and I really appreciate these updates! -Steve |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The CEP team wishes all our crunchers a great weekend! Here’s some wisdom for those of you who are interviewing for jobs these days. Enjoy!
http://xkcd.com/1094/ |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi Steve,
we have so far characterized 2 million molecular graphs represented by 19 million geometries (to capture different conformers) in 135 million DFT calcs. It's not clear yet how these numbers will develop in the coming years as we will go back to our best candidates from this first screening and do additional studies of other properties and aspects. In parallel we are developing the next generations of candidate libraries based on the experience with our primary library. Best wishes from Your Harvard CEP team |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
So it looks like we may be propagating the Backblaze design we are using in our Jabba data storage pod around the Harvard campus. There is now some interest in benchmarking this technology for high-performance applications. Let’s see what will come out of this.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
So it looks like we may be propagating the Backblaze design we are using in our Jabba data storage pod around the Harvard campus. There is now some interest in benchmarking this technology for high-performance applications. Let’s see what will come out of this. Wouldn't hurt my feelings if you posted physical failure rates with the drives you use - while noting that it is often better to phrase such things as "Have had good results with (capacity/model #) drives of manufactured by vendor 'X'...". Such information might be useful to those who are looking at dedicating multi-core boxes to CEP2, which I've often thought could be/should be used for burn-in/proof-of-concept testing by computer manufacturers. (Two birds, one stone.)(Me, I'm sorely tempted to mention a certain vendor whose spectacular failure rates under CEP2's workloads leave me thinking that they should be producing hard drives for the paperweight market - only.  Although in all fairness, it was a new line with new throughput capabilities that I suspect was simply not ready for prime time.)[Edit 1 times, last edit by Former Member at Dec 10, 2012 12:59:42 PM] |
|||
|
||||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi ibsteve2u,
Sure, happy to provide this info, in particular since we made very good experience overall. In the beginning we used 2TB Seagate drives which showed some compatibility issues with the overall Jabba design, so we cannot really comment on their reliability. Since then switched to 3TB Hitachi Deskstar 5k3000 and from the 150 drives in service so far we had (I think) one failure and one drive that was defect out of the box. The older Hitachis have been spinning for 2 years now and the most recent ones for a couple of days. I think that's a pretty good statistic. The biggest problem with the Hitachis was their limited availability after a storm flooded the factory in Thailand - they very very hard to get and expensive for quite a while afterwards. Best wishes Your Harvard CEP team |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
One failure out of 150...hmmm...somewhat better than I've managed. Ever. Quite the recommendation for Hitachi...
Matter of fact, I'm not even going to talk about some of the failure rates I've had, because then I'd have to go into how I really do use battery backup, and air conditioning, and watch ambient temperatures and case temperatures, and don't live on a fault line or above a subway or next to an elevated train or a quarry where they use dynamite all of the time or play hockey or soccer in between the computers or... |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
We had our monthly phone conference with our friends at IBM/WCG today and the project seems to be in good shape! Crunching on, crunching on...
 |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi ibsteve2u,
Yes, when we started this Jabba business we also expected a significantly higher failure rate and were pleasantly surprised. Best wishes from Your Harvard CEP team |
||
|
|
|