All tasks issued so far have been for the GPU ( BRP3cuda32) - so you still haven't answered the original question about not getting CPU work.
Of the two things you changed at once, it is far more likely to have been the 'Use GPU' option which triggered the work fetch, rather than the 200 Resource Share.
So, we presume that there is still a 'Don't use CPU' option effective somewhere, most likely within your Einstein@Home preferences.
It could be "Use CPU Enforced by version 6.10+" as Gundolf mentioned, or it could be a mix-up in "Run only the selected applications" (towards the bottom). I'd recommend that you have 'Global Correlations S5 HF search #1' ticked - note that's the "HF" version, current now but maybe not the last time you were here.
I did answer, see message 111042. I'm using 8 CPU's. This doesn't copy too well here, sorry so have edited....
Resource share 200 --- --- ---
Use CPU yes --- --- ---
Use NVIDIA GPU yes --- --- ---
Graphics setting: frames per second (FPS) 20 -- -- --
Graphics setting: render quality medium -- -- --
Graphics setting: window width (pixels) 800 -- -- --
Graphics setting: window height (pixels) 600 -- -- --
Run only the selected applications (all applications) -- -- --
Run CPU versions of applications for which GPU ... yes -- -- --
Edit
You have a BUNCH of GPU units to crunch right now! I see you crunch for other cpu projects too, do they all have work? If so you could 'owe' them cpu time and Einstein is just on the back burner for right now. This is something Boinc does automatically, though not always efficiently in the short term, to try and manage multiple projects running on one machine. Try this, go into the Boinc Manager, down by the clock, and click on the projects tab and then Einstein, then click on properties on the left and then on the right side scroll down near the bottom and tell me what the 5 scheduling line numbers are. Just the numbers are okay, I can figure out where they fit in. In fact do that for each project on the machine please, as the numbers will be different for each project.
Yes I have 10 projects and they all have work except LHC@Home which is sporadic at best. They all manage to get their work done by deadline except the occasional Cosmology and that project has a lot of people complaining about short deadlines according to their message boards.
As Einstein downloaded so many WU's I've altered its resource share back to 100. Good idea or not?
Boinc has not seen fit to start any of them yet and I see their deadline is 17 Feb.
The two "work fetch priority" lines are the key - minus 799,171 for CPU, minus 125,422 for GPU.
That's what used to be called "Long Term Debt", denominated in seconds. In the ordinary course of events, BOINC won't request work from Einstein until those values get down to 1 day - 86,400 seconds. You must have done one heck of a lot of work for Einstein in the past.
There are ways of hurrying things along, if you're prepared to edit a rather delicate configuration file, but unless you're desparate to catch a gravity wave or find a pulsar, it might be best just to let the debt decay naturally.
The two "work fetch priority" lines are the key - minus 799,171 for CPU, minus 125,422 for GPU.
That's what used to be called "Long Term Debt", denominated in seconds. In the ordinary course of events, BOINC won't request work from Einstein until those values get down to 1 day - 86,400 seconds. You must have done one heck of a lot of work for Einstein in the past.
There are ways of hurrying things along, if you're prepared to edit a rather delicate configuration file, but unless you're desparate to catch a gravity wave or find a pulsar, it might be best just to let the debt decay naturally.
LOL...I think I'll wait and see, thanks for all your help.
I've been observing how BOINC works and it really needs to have some improvement in the area of exactly when the 'Running - High Priority' flag kicks in. It seems to me that starting with that status only hours prior to a deadline is not doing much good except perhaps admitting it failed in its task.
I often have to hit the 'No New Tasks' button in order to get BOINC to clear out the backlog. Perhaps I shouldn't have so many projects on the go.
I've been observing how BOINC works and it really needs to have some improvement in the area of exactly when the 'Running - High Priority' flag kicks in. It seems to me that starting with that status only hours prior to a deadline is not doing much good except perhaps admitting it failed in its task.
I often have to hit the 'No New Tasks' button in order to get BOINC to clear out the backlog. Perhaps I shouldn't have so many projects on the go.
Or you could reduce your cache size to only 1/2 a day or so. That would increase your communications to each project but as long as your internet can handle it, it would make for fewer 'high priority' crunching.
And I think the waiting for Boinc to sort it out is a good idea in your case, I also think you need more pc's so you can spread out the work a little bit. You don't need fewer projects, you just need more pc's!!
All tasks issued so far have
)
All tasks issued so far have been for the GPU ( BRP3cuda32) - so you still haven't answered the original question about not getting CPU work.
Of the two things you changed at once, it is far more likely to have been the 'Use GPU' option which triggered the work fetch, rather than the 200 Resource Share.
So, we presume that there is still a 'Don't use CPU' option effective somewhere, most likely within your Einstein@Home preferences.
It could be "Use CPU Enforced by version 6.10+" as Gundolf mentioned, or it could be a mix-up in "Run only the selected applications" (towards the bottom). I'd recommend that you have 'Global Correlations S5 HF search #1' ticked - note that's the "HF" version, current now but maybe not the last time you were here.
I did answer, see message
)
I did answer, see message 111042. I'm using 8 CPU's. This doesn't copy too well here, sorry so have edited....
Resource share 200 --- --- ---
Use CPU yes --- --- ---
Use NVIDIA GPU yes --- --- ---
Graphics setting: frames per second (FPS) 20 -- -- --
Graphics setting: render quality medium -- -- --
Graphics setting: window width (pixels) 800 -- -- --
Graphics setting: window height (pixels) 600 -- -- --
Run only the selected applications (all applications) -- -- --
Run CPU versions of applications for which GPU ... yes -- -- --
Edit
Peter
Toronto, Canada
I should add that, despite my
)
I should add that, despite my settings here, I'm not getting email responses from these forums.
Thanks for all the help ;-)
Peter
Toronto, Canada
RE: I did answer, see
)
You have a BUNCH of GPU units to crunch right now! I see you crunch for other cpu projects too, do they all have work? If so you could 'owe' them cpu time and Einstein is just on the back burner for right now. This is something Boinc does automatically, though not always efficiently in the short term, to try and manage multiple projects running on one machine. Try this, go into the Boinc Manager, down by the clock, and click on the projects tab and then Einstein, then click on properties on the left and then on the right side scroll down near the bottom and tell me what the 5 scheduling line numbers are. Just the numbers are okay, I can figure out where they fit in. In fact do that for each project on the machine please, as the numbers will be different for each project.
Yes I have 10 projects and
)
Yes I have 10 projects and they all have work except LHC@Home which is sporadic at best. They all manage to get their work done by deadline except the occasional Cosmology and that project has a lot of people complaining about short deadlines according to their message boards.
As Einstein downloaded so many WU's I've altered its resource share back to 100. Good idea or not?
Boinc has not seen fit to start any of them yet and I see their deadline is 17 Feb.
Easier for me to post a picture, sorry....;-)
I'll post separately for the other projects.
Peter
Toronto, Canada
The two "work fetch priority"
)
The two "work fetch priority" lines are the key - minus 799,171 for CPU, minus 125,422 for GPU.
That's what used to be called "Long Term Debt", denominated in seconds. In the ordinary course of events, BOINC won't request work from Einstein until those values get down to 1 day - 86,400 seconds. You must have done one heck of a lot of work for Einstein in the past.
There are ways of hurrying things along, if you're prepared to edit a rather delicate configuration file, but unless you're desparate to catch a gravity wave or find a pulsar, it might be best just to let the debt decay naturally.
OK here goes with the
)
OK here goes with the others...
Rosetta
CPDN
LHC
Milkyway
SETI
Spinhenge
Cosmology
Malaria
WCG
Peter
Toronto, Canada
RE: The two "work fetch
)
LOL...I think I'll wait and see, thanks for all your help.
Peter
Toronto, Canada
I've been observing how BOINC
)
I've been observing how BOINC works and it really needs to have some improvement in the area of exactly when the 'Running - High Priority' flag kicks in. It seems to me that starting with that status only hours prior to a deadline is not doing much good except perhaps admitting it failed in its task.
I often have to hit the 'No New Tasks' button in order to get BOINC to clear out the backlog. Perhaps I shouldn't have so many projects on the go.
Peter
Toronto, Canada
RE: I've been observing how
)
Or you could reduce your cache size to only 1/2 a day or so. That would increase your communications to each project but as long as your internet can handle it, it would make for fewer 'high priority' crunching.
And I think the waiting for Boinc to sort it out is a good idea in your case, I also think you need more pc's so you can spread out the work a little bit. You don't need fewer projects, you just need more pc's!!