Message boards : Number crunching : BOINC Pentathlon
Author | Message |
---|---|
John P. Myers Send message Joined: 13 Apr 10 Posts: 5 Credit: 860,221 RAC: 0 |
From May 13th 8PM EDT until May 18th 8PM EDT, the 5th leg of the BOINC Pentathlon will be crunching Rosetta mercilously (we nearly crashed Einstein in the first leg of this race) as the final part of this worldwide race (we are currently running Milkyway and FreeHAL. Collatz is next, then Rosetta) :D You must be on a team to participate in this event. Everyone is welcome to join SETI.USA and you may view full details of this event on our forums at www.setiusa.net My current best estimate is that Rosetta will have about 44,000,000 cobblestones of computation done during these 5 days by the 31 teams involved in this event alone. We apologize for any server hiccups that may occur :) |
Murasaki Send message Joined: 20 Apr 06 Posts: 303 Credit: 511,418 RAC: 0 |
This is great news. Did you plan to hit Rosetta during the CASP9 event, or is it purely coincidental? |
John P. Myers Send message Joined: 13 Apr 10 Posts: 5 Credit: 860,221 RAC: 0 |
This is great news. Did you plan to hit Rosetta during the CASP9 event, or is it purely coincidental? It was half planned, half luck. Before the Pentathlon began at 8PM EDT May 4th (it doesn't end until 8PM EDT May 18th so join up!), all participating teams voted on which 5 projects (one from each category) we were going to crunch. The Biology & Medicine category really came down to being between Rosetta and Docking. One of the criteria that had to be met in order for a project to be nominated was that it must be known for stability and have plenty of work available for everyone. Because of CASP9, Rosetta was an easy choice for many :) Rosetta received 11 votes while Docking only obtained 4. Also, it should be mentioned that bunkering/hoarding/stockpiling completed WUs is an allowed strategy for this event. Only credits received during the times mentioned in the first post count, so there will be a few hundred people who will perform a massive dump of completed WUs during the night of May 13th. |
Murasaki Send message Joined: 20 Apr 06 Posts: 303 Credit: 511,418 RAC: 0 |
I just spotted we are up to 131 TeraFLOPS. ^_^ |
Chilean Send message Joined: 16 Oct 05 Posts: 711 Credit: 26,694,507 RAC: 0 |
|
Chilean Send message Joined: 16 Oct 05 Posts: 711 Credit: 26,694,507 RAC: 0 |
|
Sid Celery Send message Joined: 11 Feb 08 Posts: 2125 Credit: 41,246,824 RAC: 9,126 |
Also, it should be mentioned that bunkering/hoarding/stockpiling completed WUs is an allowed strategy for this event. Only credits received during the times mentioned in the first post count, so there will be a few hundred people who will perform a massive dump of completed WUs during the night of May 13th. Please note the message here and balance of the thread. If some WUs are not returned within 18 hours of being downloaded (see thread for details) while credit is awarded, the results may not be usable. This is in spite of the deadline shown. |
John P. Myers Send message Joined: 13 Apr 10 Posts: 5 Credit: 860,221 RAC: 0 |
:D |
Max DesGeorges Send message Joined: 1 Oct 05 Posts: 35 Credit: 942,527 RAC: 0 |
Please note the message here and balance of the thread. If some WUs are not returned within 18 hours of being downloaded (see thread for details) while credit is awarded, the results may not be usable. This is in spite of the deadline shown. This is a really strange request. How many users will follow this suggestion? Many even don't know where the preferences are in the BOINC manager and only a few read this forum. It wouldn't be better for the administrators to make specific default settings for the casp9 WUs? For example a deadline of 18 hours and a runtime of 12 hours? Is that possible? |
Sid Celery Send message Joined: 11 Feb 08 Posts: 2125 Credit: 41,246,824 RAC: 9,126 |
Please note the message here and balance of the thread. If some WUs are not returned within 18 hours of being downloaded (see thread for details) while credit is awarded, the results may not be usable. This is in spite of the deadline shown. I agree, and it should be quite possible from what little I know. Whether it happens is another question entirely... :( Until then, if it's possiblepractical to make manual adjustments to our own settings then it seems we should. |
rochester new york Send message Joined: 2 Jul 06 Posts: 2842 Credit: 2,020,043 RAC: 0 |
if not usable are they recycled until they are usable? Please note the message here and balance of the thread. If some WUs are not returned within 18 hours of being downloaded (see thread for details) while credit is awarded, the results may not be usable. This is in spite of the deadline shown. |
Murasaki Send message Joined: 20 Apr 06 Posts: 303 Credit: 511,418 RAC: 0 |
The WUs with the short deadline relate to a set of CASP targets with a 3 day deadline. Because of the timescales of loading the targets into the system and analysing the results that come back that leaves only 18 hours to receive, process and return the WU. Once 18 hours is passed the chances of your results affecting the CASP submission is minimal, though I suppose they may help to clarify for the researchers whether Bakerlab submitted a good answer or not. For the majority of WUs though there is a second CASP deadline of a few weeks, leaving plenty of time to process those WUs. |
Sid Celery Send message Joined: 11 Feb 08 Posts: 2125 Credit: 41,246,824 RAC: 9,126 |
If not usable are they recycled until they are usable? The short version of what Murasaki wrote (which I agree with) is "No". On the plus side, default buffer and runtime settings fall within the 18 hour window, so it's only those who have increased these (and know how to do it) who may have a problem. |
Mod.Sense Volunteer moderator Send message Joined: 22 Aug 06 Posts: 4018 Credit: 0 RAC: 0 |
Keep in mind that all of the results are "useful" even if not "usable" within the CASP deadlines. After CASP is over and the team has a chance to breath, they drill down into the returned data and assess how they can make better predictions next time. The same way as is done with the non-CASP tasks. Yes, work units that miss short deadlines are reissued by the BOINC server ("recycled"). However, by that time the CASP deadline has passed as well and so the outcome will arrive too late to be incorporated into the predictions submitted to CASP (i.e. not "usable", but still "useful"). Rosetta Moderator: Mod.Sense |
John P. Myers Send message Joined: 13 Apr 10 Posts: 5 Credit: 860,221 RAC: 0 |
Would just like to point out that SETI.USA will hit the 100,000,000 credit milestone with Rosetta sometime early in the morning. We are quite proud of this accomplishment and welcome anyone and everyone to join our team :) |
Max DesGeorges Send message Joined: 1 Oct 05 Posts: 35 Credit: 942,527 RAC: 0 |
BOINC Pentathlon, one year later. For project administrators: expect an increase in throughput between May 11 and 16. :) Like one year ago, the teraflops should rise quite a lot, be prepared! |
rochester new york Send message Joined: 2 Jul 06 Posts: 2842 Credit: 2,020,043 RAC: 0 |
BOINC Pentathlon, one year later. sounds like good news |
rilian Send message Joined: 16 Jun 07 Posts: 23 Credit: 2,860,224 RAC: 5,333 |
BOINC Pentathlon 2012 rules http://www.seti-germany.de/boinc_pentathlon/ i crunch for Ukraine. Join our team forums about Rosetta@home |
Mod.Sense Volunteer moderator Send message Joined: 22 Aug 06 Posts: 4018 Credit: 0 RAC: 0 |
We may have to add a "Like" button :) 1.1M WUs in progress! Rosetta Moderator: Mod.Sense |
Chilean Send message Joined: 16 Oct 05 Posts: 711 Credit: 26,694,507 RAC: 0 |
|
Message boards :
Number crunching :
BOINC Pentathlon
©2024 University of Washington
https://www.bakerlab.org