Message boards :
Number crunching :
Why can't I get more Work units?
Message board moderation
Author | Message |
---|---|
Send message Joined: 10 Aug 08 Posts: 218 Credit: 41,846,854 RAC: 0 |
I'm sitting here with 2 computers that are running just fine on CPUs. I usually turn a WU over on one computer in 1.75 hours and 3.1 hours on the other. And I have to beg for WU's 1 or 2 at a time. And then I look at ones I have pending credit for and these systems are running gfx cards with nvidia and ati and have 300 to 800 WUs and they're timing out on past work and still getting more every freaking day !!!!!!! And on top of it, they aren't completing any of the work they're getting currently. Here's a couple of examples : Here Here How about opening the pipeline here so I can get at least a day's work that I can turn it without having to beg for WU's every few hours and shutting off the pipeline to those that have large numbers of uncompleted WU's so that others that can do the work can get more. |
Send message Joined: 8 Aug 12 Posts: 9 Credit: 156,273 RAC: 0 |
I seem to have something of the same problem. I got a bunch of WU's immediately upon signing up on the 8th. And I haven't received any since I finished them up rather quickly as they all jumped to high priority on my 8 CPU, 16 gig U320 15K SCSI system. I have tried hitting update a few times to get more WU's but nothing. James Joined MilkyWay@Home in 2012 Online since ArpNET days First activity on Honeywell 1648 Series Mainframe in 1975 at age 12. |
Send message Joined: 8 May 09 Posts: 3315 Credit: 519,944,632 RAC: 22,383 |
I seem to have something of the same problem. I got a bunch of WU's immediately upon signing up on the 8th. And I haven't received any since I finished them up rather quickly as they all jumped to high priority on my 8 CPU, 16 gig U320 15K SCSI system. What are your cache settings? In version 7.x they are more literal than in previous versions. |
Send message Joined: 8 May 09 Posts: 3315 Credit: 519,944,632 RAC: 22,383 |
I'm sitting here with 2 computers that are running just fine on CPUs. I usually turn a WU over on one computer in 1.75 hours and 3.1 hours on the other. And I have to beg for WU's 1 or 2 at a time. I think those too might be cache settings, except this time set too high. |
Send message Joined: 8 Aug 12 Posts: 9 Credit: 156,273 RAC: 0 |
Well since I've done SETI for 11 years I've always got it set to give me up to 10 extra days worth of work since I let all the machines in the office run 24x7 and split their CPU time between their p reprogrammed takes and BOINC. James Joined MilkyWay@Home in 2012 Online since ArpNET days First activity on Honeywell 1648 Series Mainframe in 1975 at age 12. |
Send message Joined: 8 May 09 Posts: 3315 Credit: 519,944,632 RAC: 22,383 |
Well since I've done SETI for 11 years I've always got it set to give me up to 10 extra days worth of work since I let all the machines in the office run 24x7 and split their CPU time between their p reprogrammed takes and BOINC. But in version 7 the cache settings changed and the meaning are more LITERAL now, as in "minimum work buffer" now means do not get anymore work until Boinc reaches this time frame. The second setting "max additional work" means do not get more than this amount of work over and above the first number. So when I have my first number at 0.75 days and the second number at 0.10 days I end up with a cache of just under 1 days worth of work. So I will ask again did you change the settings when going from version 6 to version 7? The PROGRAM changed, so if YOU did not change the settings they do not now reflect what YOU want! The Boinc program has changed MANY times in the past 11 years as you well know, it has changed AGAIN with the version 7 series. On a side note I see your Seti stats show a 4k rac, do you use your gpu there? |
Send message Joined: 8 Aug 12 Posts: 9 Credit: 156,273 RAC: 0 |
I redid all my settings for both computing preferences and project preferences when I added the additional projects. I've been on version 7 for awhile. Although 1 workstation is on version 6 because its OS literally can't go higher than the version its running. I originally had the first value as "--" and the second as "10" I went and changed everything to "2" and kept the "10". I have GPU checked for use always although I'm not sure watching BOINC how well it performs TSO operations. And from what I can see my GPU runs under a constantly fluctuating load from 40% to 80% (I run 5 extra fans to cool the CPU's and the GPU). I've got my processor settings at 75% of the CPU's at 75% load. So I get 6 primary tasks and 1 sub task running all at once. Now that I've said that I dont know if that impacts on your question. Is the figure you were mentioning impacted by multiple machines on 1 account? Because several of the machines don't have GPU they can use as they are older machines. And my primary workstation apparently has 1 card below the recommended list. So last night I ordered the highest card my MB will take which is on the list. James Joined MilkyWay@Home in 2012 Online since ArpNET days First activity on Honeywell 1648 Series Mainframe in 1975 at age 12. |
Send message Joined: 8 May 09 Posts: 3315 Credit: 519,944,632 RAC: 22,383 |
I redid all my settings for both computing preferences and project preferences when I added the additional projects. I've been on version 7 for awhile. Although 1 workstation is on version 6 because its OS literally can't go higher than the version its running. If you want to max out your work units, there is a sub limit here on MW, then put 10 in the first box and a 2 in the second box for the cache settings. As I said ver 7 is LITERAL, so if you have a 2 in the first box it will let your cache drop down to only 2 days worth of units before it asks for more, then it will get 10 days worth to fill up to the second number. As for your gpu it does not seem to be crunching as you have no tasks. You will need to go to Nvidia and get the new drivers from them to let MW use your gpu. The standard Microsoft drivers let you play minor games and view the screen but the real upgrades come from the manufacturers drivers and that is what MW, Seti etc, etc NEED to crunch with. I have several gpu's I COULD, and have in the past, put here at MW but am using them to crunch for Moo right now. I did not get to 9.5 million with just one gpu. So yes the settings WILL work for multiple machines. One thing I do is to set up each machine on its own thru the Boinc icon down by the clock. Most of the settings are available thru it and it is on a machine by machine basis. NEW machines will adopt older machines settings in some cases, so you have to be careful when you add machines. I currently have 15 pc's crunching for Boinc over several different projects, most of them have gpu's crunching in them. I will let my signature come thru this time, I usually on do that on Sunday, so you can get an idea of what I am crunching for right now. ps I JUST noticed my RAC is getting low here again so I will crunch for today for MW on one of my pc's. When the RAC gets too low you can't post here anymore. Have you checked out the Cafe Milkyway yet? There is a LPTPWT, last person to post wins thread, it is just about having fun and a quick post saying you are winning. The thread starter determines who wins based on whatever they chose, then the winner gets to start the next thread. |
Send message Joined: 19 Jul 10 Posts: 578 Credit: 18,845,239 RAC: 856 |
As for your gpu it does not seem to be crunching as you have no tasks. You will need to go to Nvidia and get the new drivers from them to let MW use your gpu. The standard Microsoft drivers let you play minor games and view the screen but the real upgrades come from the manufacturers drivers and that is what MW, Seti etc, etc NEED to crunch with. He has version 301.42 drivers installed, that's OK. The real problem is, he has a GeForce GTS 250: no DP -> no Milkyway. See also the GPU Requirements Thread, he got there already that answer. |
©2024 Astroinformatics Group