work drops when BOINC program closed |
![]() |
| log in |
Questions and Answers : Preferences : work drops when BOINC program closed
| Author | Message |
|---|---|
|
I have been running two projects at 100% with dual core processor and BOINC program exited. That worked fine until two weeks ago. Now my production appears to have stopped. I notice that my message file appears to start over each time I enter BOINC program. That is new. I am not running the screen saver, but have the client running 100% cpu in the background 24/7. What changed? I cannot seem to set any preference to get this to keep working. | |
| ID: 407255 · | |
I have been running two projects at 100% with dual core processor and BOINC program exited. That worked fine until two weeks ago. Now my production appears to have stopped. I notice that my message file appears to start over each time I enter BOINC program. That is new. I am not running the screen saver, but have the client running 100% cpu in the background 24/7. What changed? I cannot seem to set any preference to get this to keep working. By exited do you mean minimized to the taskbar? or do you mean you actually exited the program? Do you mean you have 2 instances of Boinc running, each doing a different project? or do you mean you have a dual core/HT machine doing 2 units at once and switching between 2 projects? ____________ | |
| ID: 407303 · | |
|
Goto preferences and set leave in memory to YES. | |
| ID: 407352 · | |
|
I did have the Seti job set to not stay in memory. I changed that to a YES, | |
| ID: 407434 · | |
I did have the Seti job set to not stay in memory. I changed that to a YES, What do you mean by "dropping""" ____________ And the beat goes on Sonny and Cher BOINC Wiki | |
| ID: 407467 · | |
|
Did you upgrade the software a couple of weeks ago? Did you install it differently than before. | |
| ID: 407539 · | |
I did have the Seti job set to not stay in memory. I changed that to a YES, By 2 jobs you do not mean 2 instances of Boinc, correct? You mean 2 different workunits? If the later is correct, then your computer is doing as it should do. The manager controls Boinc and should be giving 100% of each workunit to each processor. These will change depending on the settings you have set for how much time to put into each project you run. For isntance if you want 90% on seti and 10% on Einstein then periodically the manager will switch your computer so over the long run 90% of the time your ocmputer will crunch for Seti and 10% od the time it will crunch for Einstein. The short term numbers will not be accurate!! It is not a perfect piece of software, but over the long run it will come out fine. In the short term you will be off by a couple of percentage points one way or the other. ____________ | |
| ID: 408155 · | |
|
I removed BOINC from my computer, went to the setiathome web site and downloaded and installed BOINC again. It appears to be working now. | |
| ID: 408187 · | |
I removed BOINC from my computer, went to the setiathome web site and downloaded and installed BOINC again. It appears to be working now. This could be part of your problem, you cannot do 100% for more than 1 project when you are splitting your time between multiple projects. You only have 100% of crunching time available, if you want each project to get 50% of your time, you need to put that. You cannot have one project run on one cpu and another project run on the other cpu, right now. Boinc will physically stop crunching on one project and switch to the other based on the numbers you put in. If you want Climate and Seti to each have 50% of your crunching time than you should put 50/50 on the line. Use other numbers if you want something else, but do not exceed 100 for the total. ____________ | |
| ID: 411406 · | |
This could be part of your problem, you cannot do 100% for more than 1 project when you are splitting your time between multiple projects. You only have 100% of crunching time available, if you want each project to get 50% of your time, you need to put that. You cannot have one project run on one cpu and another project run on the other cpu, right now. Boinc will physically stop crunching on one project and switch to the other based on the numbers you put in. If you want Climate and Seti to each have 50% of your crunching time than you should put 50/50 on the line. Use other numbers if you want something else, but do not exceed 100 for the total. Not so. If it makes it easier to think about them to make the total 100, and if your hosts (if more than one) are all attached to the same set of projects, that’s fine, but it’s not at all necessary. The resource-share numbers are not percentages, but ‘weights’. BOINC calculates the percentages from those, according to the projects that host is attached to. For example, at the moment my “home†(and default) settings are: 100 Einstein@home 100 Pirates@home 100 Seti@home 175 Seti@home Beta 100 SZTAKI Desktop Grid. Note that the only project with anything but the default 100 share is S@h Beta: that’s the only site where I had to edit the preferences, the others being unchanged. (Not to say I haven’t actually twiddled them before, but had I settled on this allocation before I first started, I could have left them all alone.) My Mac G5 is attached to the first four projects above, so E@h, P@h, & S@h are getting 100/475 = 21% of its CPU time each, while S@h Beta gets 175/475 = 37%. Note that the total is 475, not 575: as far as this machine is concerned, SDG might as well not exist (and its BOINC Manager’s user-total graph doesn’t include any of the credit I’ve earned there). My G4/733 at work—but still in the “home†venue—is attached to all the projects except Pirates, so its numbers are the same as the G5’s, substituting SDG for P@h. OTOH one of the G4/400s there is only attached to E@h and S@h, so it runs 100/200 = 50% of each. Then there’s the WinXP/AMD system that actually is in the “work†venue, where E@h has its share set to 300 (and BOINC is set to stop when the system’s in use). That machine is attached to S@h and SDG as well (I have no “work†venue preferences at all on those projects, so they really are untouched): they get 100/500 = 20% of its time each, while E@h gets 300/500 = 60%. Anyway, my goal is to minimize the number of places I have to change preferences in order to reallocate my resources, but if you find it easier to deal with percentages I’m sure you’re not alone, and whatever works for you is best. However, if you consider that BOINC has to handle users like me, with varying combinations of hosts and projects, I think you’ll see that a simple percentage system wouldn’t work. Something else to think about: if you had to keep your total at 100, what would happen if you changed the preferences at one or more projects, but then found that another was having an outage, so you couldn’t get in to make it match? It would be unreasonable to expect you to dash back through all your projects and reverse the changes before one of your hosts got an update and stopped with a “total not 100†error! Every BOINC function must be ‘robust’ WRT lack of availability of updates from any or all projects. ____________ | |
| ID: 411622 · | |
Questions and Answers : Preferences : work drops when BOINC program closed
| Copyright © 2013 University of California |