| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: FightAIDS@Home Phase 2
Thread: 24 Hour Deadline too Short |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 112
|
|
| Author |
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges: 
|
Out of curiosity how long did the task run for & are you a Linux or Windows user They run for 4 hours 15 minutes on my i7-4771 (Windows 7 64-bit), running on six out of the eight cores, with the other two cores free. So they would take longer per work unit if running on all eight cores, around 5 hours 20 minutes. That is at the default clock of 3.7 GHz. They take somewhat longer on Linux, as I have posted elsewhere. Also, I set my buffer to 0.1 + 0.4 days, and don't have any high priorities. [Edit 1 times, last edit by Jim1348 at Jun 24, 2019 2:23:06 PM] |
||
|
|
Speedy51
Veteran Cruncher New Zealand Joined: Nov 4, 2005 Post Count: 1261 Status: Offline Project Badges: 
|
Thanks Jim for your input. Timing though provided in my previous post was running and 16 cores. It's interesting it sounds like your machine gets the same time roughly as I got, I'm using a Haswell E I see your using Haswell. Apologies everybody for the off topic conversation
---------------------------------------- |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges:
|
It's interesting it sounds like your machine gets the same time roughly as I got, I'm using a Haswell E I see your using Haswell. Even my Coffee Lake (i7-8700) usually does not get much more per core, it just has more cores. They are all basically glorified Haswells. |
||
|
|
Robokapp
Senior Cruncher Joined: Feb 6, 2012 Post Count: 248 Status: Offline Project Badges:
|
I'd like to ask a silly yet scary question:
I am thinking about "fishing" the 10-year badge since I'm at 9 years. but if I have my computers pick up a 2-day queue and the WUs only last 24 hours does that mean my computer will sh*t the bed and end up running nothing but expired units? or will it intelligently know not to bite more that it can chew? do I need to manually manage it or will BOINC do it for me? ultimately i have no problem doing it by hand, this is more of a 'how do magnets work" question. |
||
|
|
adriverhoef
Master Cruncher The Netherlands Joined: Apr 3, 2009 Post Count: 2083 Status: Offline Project Badges: 
|
I am thinking about "fishing" the 10-year badge since I'm at 9 years. but if I have my computers pick up a 2-day queue and the WUs only last 24 hours does that mean my computer will sh*t the bed and end up running nothing but expired units? or will it intelligently know not to bite more that it can chew? do I need to manually manage it or will BOINC do it for me? ultimately i have no problem doing it by hand, this is more of a 'how do magnets work" question. If the deadline passes for a task that hasn't started yet, BOINC will abort that task and a new copy will be sent to another volunteer (a so-called wingman). BOINC tries to maximize the number of tasks it can run within a certain time (before a certain deadline). So I think it will be very hard to load, let's say, 10 days worth of FAH2 tasks. However, if it's still too much, you could limit the number of FAH2 tasks in your device profile: go to the Device Manager (https://www.worldcommunitygrid.org/ms/device/viewDevices.do), select your Device Profile, tick Custom Profile for advanced options and settings, then find Project Limits where you can adjust the maximum number of tasks assigned to your device for FAH2. [Edit 2 times, last edit by adriverhoef at Jul 16, 2019 2:08:20 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I have a 3 day queue specified but can only get 1 FAH2 unit per core/thread. They have purposely set it up that way.
|
||
|
|
rbotterb
Senior Cruncher United States Joined: Jul 21, 2005 Post Count: 401 Status: Offline Project Badges:
|
Not sure how this is working, but lately almost all my WUs from this project are finishing up about 2-3 hours past the 24 hr deadline, yet they are coming up valid eventually. SO is someone looking at these longer WUs manually and flipping the switch to Valid? If this is occurring regularly, I don't understand why the researchers don't have a longer period on their deadline. Now if they don't want a week or 10 days for some reason, would it hurt much to go to 48 hours?
|
||
|
|
Robokapp
Senior Cruncher Joined: Feb 6, 2012 Post Count: 248 Status: Offline Project Badges:
|
it is odd...it is also awkward to work with. But maybe it's for a reason??
|
||
|
|
katoda
Senior Cruncher Poland Joined: Apr 28, 2007 Post Count: 170 Status: Offline Project Badges:
|
SO is someone looking at these longer WUs manually and flipping the switch to Valid? Nope, there is no manual intervention. It works in the following way: 1. you receive WU with 24 hours of deadline 2. deadline is passing, you did not complete your WU. It receives status "no reply" and a second copy is issued to another device in WCG network. 3. now is the tricky part: if you manage to complete calculations before the second WU copy is completed and returned, you will receive the full credit and your WU will be marked as valid. If not - the second copy will be marked as valid and yours will become "Too late" without any credit. AFAIK the system will abort the second copy if the device did not start computing it. If the computations already started, it will receive credit as well, but I'm not 100% sure about this part.  |
||
|
|
rbotterb
Senior Cruncher United States Joined: Jul 21, 2005 Post Count: 401 Status: Offline Project Badges:
|
Thank for the update on how they process WUs on this project katoda. Makes sense what I'm seeing. I'll just let my WUs go even if past the 24 hour mark and see if I get lucky more often than not.
|
||
|
|
|