Message boards :
Number crunching :
Need Help With BOINC (6.10.56) and Resource Share 0
Message board moderation
Author | Message |
---|---|
Cruncher-American Send message Joined: 25 Mar 02 Posts: 1513 Credit: 370,893,186 RAC: 340 |
I am running SETI along with Collatz (to keep my GPUs warm). SETI is share 700, Collatz, 0. I have about 200 SETI CUDA WUs I got last night (good), and a few dozen Collatz WUs. I find that even though Collatz is 0, I am still running those WUs instead of SETI. I thought (based on some discussion back in these threads when SETI went down for the new servers) that the 0 for Collatz implied that when I got some SETI WUs, they would run before any Collatz WUs were started. So what's wrong? Did I misinterpret what was said? Incidentally, on my other machine, which is running Einstein (share 0) and SETI (share 700), I got some SETI MB WUs, and they are running in tandem on that machine - it's an 8-core system, and it is running 4 Einstein and 4 SETI simultaneously...also on 6.10.56. How is that consistent with the above???? Thanks for any info.... |
arkayn Send message Joined: 14 May 99 Posts: 4438 Credit: 55,006,323 RAC: 0 |
|
John McLeod VII Send message Joined: 15 Jul 99 Posts: 24806 Credit: 790,712 RAC: 0 |
Also, once the tasks are downloaded to the machine, they have to be completed by the deadline. The resource fraction is used in the simulator to determine how much wall time they will have before they complete. Since the resource fraction is 0 (0 / (0 + 700)), they will run high priority until completed. BOINC WIKI |
BilBg Send message Joined: 27 May 07 Posts: 3720 Credit: 9,385,827 RAC: 0 |
I wonder what will be if I attach to 2 projects (only) - both with Resource share 0 Will BOINC divide by zero? ;) (0 / (0 + 0)) Â - ALF - "Find out what you don't do well ..... then don't do it!" :) Â |
Cruncher-American Send message Joined: 25 Mar 02 Posts: 1513 Credit: 370,893,186 RAC: 340 |
Arkayn - thanks - that at least tells me why the two cases are different... Also, once the tasks are downloaded to the machine, they have to be completed by the deadline. The resource fraction is used in the simulator to determine how much wall time they will have before they complete. Since the resource fraction is 0 (0 / (0 + 700)), they will run high priority until completed. But then this makes no sense - it says the share 0 project will always get run first, just the reverse of what it is supposed to mean. At least as was explained before in the Forums, where it was said that a non-0 share project would completely take over the system when it had work - that's why I added Einstein with 0 share, so SETI would immediately begin running as currently running Einstein WUs finished when WUs for it (SETI) returned, even though other Einstein WUs were pending. And this certainly doesn't explain why Einstein and SETI were each getting 4 of 8 CPUs. Color me confused. |
Metod, S56RKO Send message Joined: 27 Sep 02 Posts: 309 Credit: 113,221,277 RAC: 9 |
Arkayn - thanks - that at least tells me why the two cases are different... JM7 explained, perhaps with too little words. When BOINC client decides which projects to contact for new work, it takes 'resource share' quite verbatim. But when BOINC client decides which project tasks to run, it needs to consider deadlines. Example: you have tasks from two projects, one having double resource share than the other. If all tasks consume same amount of CPU time, it would mean that first project runs 2/3 of time and the other 1/3 of time. That is, when all tasks from both projects can be run with the normal CPU time split and none has possibility of missing deadline. If this possibility kicks in, then tasks of that project will be run in high priority mode (will have priority over other tasks). If one project has resource share of 0 and there are tasks for other project available, then tasks of the zero resource project would be scheduled zero CPU time. Which means they won't get done before deadline. Which triggers high priority for that project. In your case, how many Einstein tasks are on your computer? Nevertheless, BOINC won't request new tasks for the project with resource share of 0 as long there's some work for other project(s). Metod ... |
Cruncher-American Send message Joined: 25 Mar 02 Posts: 1513 Credit: 370,893,186 RAC: 340 |
Nevertheless, BOINC won't request new tasks for the project with resource share of 0 as long there's some work for other project(s). Not apparently true. What happens (over several times): the computer waits until the last WU for the share 0 project to be started, then asks for (1 days' worth) more. This happened on both machines when running Einstein (CPU only) (0), Collatz (0) and SETI (700) when SETI had no WUs to run or to download. That is, the only projects actually running were share 0, so maybe something else comes into play? |
Gundolf Jahn Send message Joined: 19 Sep 00 Posts: 3184 Credit: 446,358 RAC: 0 |
Nevertheless, BOINC won't request new tasks for the project with resource share of 0 as long there's some work for other project(s). But that's exactly what Metod, S56RKO said: You didn't have any tasks for your "above zero" project, so the cache was filled with tasks from the backup project(s). There have been requests that only one task per core is downloaded from backup projects, but the powers that be have decided to fill the cache when one resource is idle. Gruß, Gundolf Computer sind nicht alles im Leben. (Kleiner Scherz) SETI@home classic workunits 3,758 SETI@home classic CPU time 66,520 hours |
Aurora Borealis Send message Joined: 14 Jan 01 Posts: 3075 Credit: 5,631,463 RAC: 0 |
I haven't tested the feature but from reading the mailing lists, V6.12.x will only get one WU per idle CPU/GPU from 0 resource projects. The project does have to have updated their servers to actually accept 0 resource values in the preference setting. |
arkayn Send message Joined: 14 May 99 Posts: 4438 Credit: 55,006,323 RAC: 0 |
I haven't tested the feature but from reading the mailing lists, V6.12.x will only get one WU per idle CPU/GPU from 0 resource projects. The project does have to have updated their servers to actually accept 0 resource values in the preference setting. Yep, the 6.11.x and 6.12.x versions did employ the only get 1 unit per idle resource function and it does work exactly like it should. |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
I haven't tested the feature but from reading the mailing lists, V6.12.x will only get one WU per idle CPU/GPU from 0 resource projects. The project does have to have updated their servers to actually accept 0 resource values in the preference setting. No not really arkayn. I´ve set collatz at 0 share and get 149 units when CC calls for work. On Milkyway i only get 1 at a time for a 6 core it should be 36 at least. So it dont work well except here. With each crime and every kindness we birth our future. |
Gundolf Jahn Send message Joined: 19 Sep 00 Posts: 3184 Credit: 446,358 RAC: 0 |
Yep, the 6.11.x and 6.12.x versions did employ the only get 1 unit per idle resource function and it does work exactly like it should. As Aurora Borealis wrote, the project does have to have updated their servers to actually accept 0 resource values in the preference setting. (And to react accordingly.) Gruß, Gundolf Computer sind nicht alles im Leben. (Kleiner Scherz) SETI@home classic workunits 3,758 SETI@home classic CPU time 66,520 hours |
arkayn Send message Joined: 14 May 99 Posts: 4438 Credit: 55,006,323 RAC: 0 |
I haven't tested the feature but from reading the mailing lists, V6.12.x will only get one WU per idle CPU/GPU from 0 resource projects. The project does have to have updated their servers to actually accept 0 resource values in the preference setting. When I have my resource share at Collatz set to 0, I only get one unit at a time on both my Nvidia and ATI machines. Right now I have the Nvidia machine set up to normal numbers to keep the GPU warm. You are also only running 6.10.58 which does not have the fix in it. |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
That explains it. Thanks arkayn. Don´t like the 6.12.x very much but will give it another shot. With each crime and every kindness we birth our future. |
Metod, S56RKO Send message Joined: 27 Sep 02 Posts: 309 Credit: 113,221,277 RAC: 9 |
Don´t like the 6.12.x very much ... I was thinking of trying that one, so I wonder why don't you like it? Metod ... |
Gundolf Jahn Send message Joined: 19 Sep 00 Posts: 3184 Credit: 446,358 RAC: 0 |
I was thinking of trying that one, so I wonder why you don't you like it. Most people don't like that the Messages tab is missing (well, hidden). However, you can use the 6.10 manager (including Messages tab) with the 6.12 client. There've been threads on that topic. Gruß, Gundolf |
bill Send message Joined: 16 Jun 99 Posts: 861 Credit: 29,352,955 RAC: 0 |
I was thinking of trying that one, so I wonder why you don't you like it. Anyway to get rid of that stupid "Notices" tab? Or move it all the way to the right so the other tabs are back where they should be? And yes, either put messages back where it was or make "Event Log" always on top. I'm tired of chasing it down for messages. |
Jamie Send message Joined: 5 Apr 06 Posts: 162 Credit: 9,867,955 RAC: 0 |
I was thinking of trying that one, so I wonder why you don't you like it. No way that I'm aware of, but as Gandolf said, you can use the 6.10.xx manager with the 6.12.xx client It's just a case of installing the 6.12.xx app then dropping in the 6.10.xx boincmgr.exe (and possibly the boinc.dll - been a while since I last done this) file into the install directory. You then get your messages tab back as well, lose the notices tab and still get the newer client build |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
Don´t like the 6.12.x very much ... Its not the stupid message tab btw. I don´t like the settings are missing. Sometimes i turn use graphics while computer is in use sometimes. It works perfectly with MW while i´m watching the boards on my Phenom. With each crime and every kindness we birth our future. |
bill Send message Joined: 16 Jun 99 Posts: 861 Credit: 29,352,955 RAC: 0 |
OK, reinstalled 6.12.8 with the 6.10.58 client. We'll see what happens. Thanks for the pointer. |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.