I currently run 6.6.38 which is the highest available for my CUDA driver (you need a version of the driver that is not supported on my device to go to the 6.10.X version).
I know I'm joining this conversation rather late (I've only just seen it) and I don't own an NVIDIA GPU, so please excuse the dumb question.
Are you saying that you cannot install 6.10.x at all with whatever version driver you currently have for your GPU? I would think you should be able to upgrade to 6.10.x just fine seeing that you now have disabled the use of the GPU in your preferences. You need a 512MB card as a bare minimum anyway and yours shows as only 256MB.
The other thing that puzzles me is that there are currently no tasks of any sort showing in the list for hostID 986988. This implies that you haven't had any E@H work at all for quite a while (tasks hang around for at least a week after they are returned - longer if they are pending for a while). It's hard to see why E@H could be indebted to other projects at all since it doesn't seem to have done any E@H work for quite a while.
The debt figure of -543421.26 that you mention is rather large and will take quite a while to work off. You might consider zeroing your debts to speed up the process. Check out the zero_debts option under client configuration.
The new Nvidia API that BOINC 6.10 uses has a minimum driver set of CUDA 2.2, 185.85.
If your present drivers are below this version, update first.
If your present drivers are below this and they're the last available for your hardware, you cannot update to 6.10; stay at the last 6.6.x version for your OS.
Spread the word.
ca77738 is using (from the third message in this thread):
NVIDIA GPU 0: Quadro NVS 160M (driver version 17968, compute capability 1.1, 256MB, est. 4GFLOPS)
ca77738 is using (from the third message in this thread):
NVIDIA GPU 0: Quadro NVS 160M (driver version 17968, compute capability 1.1, 256MB, est. 4GFLOPS)
But if ca77738 is not intending to do GPU crunching, surely he could still upgrade to 6.10.x if he wished and just run CPU apps?
That's fine, provided he doesn't want to use the CUDA card for any BOINC project. He's attached to SETI, where 256MB is enough: and I think his v179.68 driver just scrapes under the wire, too.
ca77738 is using (from the third message in this thread):
NVIDIA GPU 0: Quadro NVS 160M (driver version 17968, compute capability 1.1, 256MB, est. 4GFLOPS)
But if ca77738 is not intending to do GPU crunching, surely he could still upgrade to 6.10.x if he wished and just run CPU apps?
That's fine, provided he doesn't want to use the CUDA card for any BOINC project. He's attached to SETI, where 256MB is enough: and I think his v179.68 driver just scrapes under the wire, too.
First - Thanks to all who have replied! I appreciate all the advice and suggestions.
I do use the CUDA for SETI work and it works well. Yes, you are right my driver and memory just barely cut the mustard but it's a laptop and I can't do much about it (unless you know of a way to put a fancy CUDA card on a PCMCIA/USB/SATA slot?).
I have tried to upgrade to 6.10 each time they do a new revision and the install goes fine but then it runs into the CUDA driver issue and I have to downgrade again. I stopped trying when I saw the above quote from Ageless for the minimum driver version and realized I couldn't make it (thanks Dell!).
Checked again this morning and I still can't download any E@H work units - same error. The balance of the work fetch has changed so that e@H is 3rd out of the 4 I run (Climate, E@H, Rosetta, SETI). I've even tried "No New Tasks" on the other projects but it still complains about needing to update the client.
I ran the zero_debts option and it cleared everything out (I removed the entry after per the instructions). Still have the same error "client is old" on a manual refresh.
Are you saying that you cannot install 6.10.x at all with whatever version driver you currently have for your GPU? I would think you should be able to upgrade to 6.10.x just fine seeing that you now have disabled the use of the GPU in your preferences. You need a 512MB card as a bare minimum anyway and yours shows as only 256MB.
I have upgraded to 6.10.X each time a new release came out hoping things would be fixed but stopped when I saw Ageless' note about driver version. It's a laptop so there isn't much more I can do.
Quote:
The other thing that puzzles me is that there are currently no tasks of any sort showing in the list for hostID 986988. This implies that you haven't had any E@H work at all for quite a while (tasks hang around for at least a week after they are returned - longer if they are pending for a while). It's hard to see why E@H could be indebted to other projects at all since it doesn't seem to have done any E@H work for quite a while.
I was sick for about two weeks earlier this month and didn't pay much attention to how things were going. According to my stats the last time I really had data from E@H was about Dec 8. Don't know why I would be in such a debt situation.
Quote:
You might consider zeroing your debts to speed up the process.
Did this. Thanks! Still complaining about my client version.
*sigh* Might have to shift a couple other slower machines from their projects to E@H to compensate.
You could get rid of the error message by unchecking the Arecibo Binary Pulsar Search work in the project preferences.
Did this. End up with:
12/24/2009 10:14:43 AM Einstein@Home update requested by user
12/24/2009 10:14:47 AM Einstein@Home Sending scheduler request: Requested by user.
12/24/2009 10:14:47 AM Einstein@Home Requesting new tasks for GPU
12/24/2009 10:14:52 AM Einstein@Home Scheduler request completed: got 0 new tasks
12/24/2009 10:14:52 AM Einstein@Home Message from server: No work sent
12/24/2009 10:14:52 AM Einstein@Home Message from server: No work is available for Hierarchical S5 all-sky GW search #5
12/24/2009 10:14:52 AM Einstein@Home Message from server: No work is available for Hierarchical S5 all-sky GW search #6
12/24/2009 10:14:52 AM Einstein@Home Message from server: No work available for the applications you have selected. Please check your settings on the web site.
Which is admittedly better then "client to old" but still no work...
Quote:
I thought I'd add my two cents, seeing how I am unceremoniously being talked about in this thread. ;-)
But we mean it in the nicest possible way! ;-) Notoriety is just the price you pay for providing that magic piece of information that everyone now refers to. Of course, in my case the info just supported what I've always thought about Dell... (the laptop was given to me so I guess I really shouldn't complain much)
RE: Based on what you said
)
At least, that's the way I understand it. :-)
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
RE: At least, that's the
)
Thanks for the input. I'll monitor it over the next day or so and see if it allows me to download.
Mike
RE: I currently run 6.6.38
)
I know I'm joining this conversation rather late (I've only just seen it) and I don't own an NVIDIA GPU, so please excuse the dumb question.
Are you saying that you cannot install 6.10.x at all with whatever version driver you currently have for your GPU? I would think you should be able to upgrade to 6.10.x just fine seeing that you now have disabled the use of the GPU in your preferences. You need a 512MB card as a bare minimum anyway and yours shows as only 256MB.
The other thing that puzzles me is that there are currently no tasks of any sort showing in the list for hostID 986988. This implies that you haven't had any E@H work at all for quite a while (tasks hang around for at least a week after they are returned - longer if they are pending for a while). It's hard to see why E@H could be indebted to other projects at all since it doesn't seem to have done any E@H work for quite a while.
The debt figure of -543421.26 that you mention is rather large and will take quite a while to work off. You might consider zeroing your debts to speed up the process. Check out the zero_debts option under client configuration.
Cheers,
Gary.
RE: Are you saying that you
)
Yes. Quoting Ageless:
ca77738 is using (from the third message in this thread):
NVIDIA GPU 0: Quadro NVS 160M (driver version 17968, compute capability 1.1, 256MB, est. 4GFLOPS)RE: ca77738 is using (from
)
But if ca77738 is not intending to do GPU crunching, surely he could still upgrade to 6.10.x if he wished and just run CPU apps?
Cheers,
Gary.
RE: RE: ca77738 is using
)
That's fine, provided he doesn't want to use the CUDA card for any BOINC project. He's attached to SETI, where 256MB is enough: and I think his v179.68 driver just scrapes under the wire, too.
RE: RE: RE: ca77738 is
)
First - Thanks to all who have replied! I appreciate all the advice and suggestions.
I do use the CUDA for SETI work and it works well. Yes, you are right my driver and memory just barely cut the mustard but it's a laptop and I can't do much about it (unless you know of a way to put a fancy CUDA card on a PCMCIA/USB/SATA slot?).
I have tried to upgrade to 6.10 each time they do a new revision and the install goes fine but then it runs into the CUDA driver issue and I have to downgrade again. I stopped trying when I saw the above quote from Ageless for the minimum driver version and realized I couldn't make it (thanks Dell!).
Checked again this morning and I still can't download any E@H work units - same error. The balance of the work fetch has changed so that e@H is 3rd out of the 4 I run (Climate, E@H, Rosetta, SETI). I've even tried "No New Tasks" on the other projects but it still complains about needing to update the client.
I ran the zero_debts option and it cleared everything out (I removed the entry after per the instructions). Still have the same error "client is old" on a manual refresh.
Any other ideas?
Again - thanks to all who have replied!
RE: Are you saying that you
)
I have upgraded to 6.10.X each time a new release came out hoping things would be fixed but stopped when I saw Ageless' note about driver version. It's a laptop so there isn't much more I can do.
I was sick for about two weeks earlier this month and didn't pay much attention to how things were going. According to my stats the last time I really had data from E@H was about Dec 8. Don't know why I would be in such a debt situation.
Did this. Thanks! Still complaining about my client version.
*sigh* Might have to shift a couple other slower machines from their projects to E@H to compensate.
You could get rid of the
)
You could get rid of the error message by unchecking the Arecibo Binary Pulsar Search work in the project preferences.
Although they run on the CPU as well (I think? I don't do them).
I thought I'd add my two cents, seeing how I am unceremoniously being talked about in this thread. ;-)
RE: You could get rid of
)
Did this. End up with:
12/24/2009 10:14:43 AM Einstein@Home update requested by user
12/24/2009 10:14:47 AM Einstein@Home Sending scheduler request: Requested by user.
12/24/2009 10:14:47 AM Einstein@Home Requesting new tasks for GPU
12/24/2009 10:14:52 AM Einstein@Home Scheduler request completed: got 0 new tasks
12/24/2009 10:14:52 AM Einstein@Home Message from server: No work sent
12/24/2009 10:14:52 AM Einstein@Home Message from server: No work is available for Hierarchical S5 all-sky GW search #5
12/24/2009 10:14:52 AM Einstein@Home Message from server: No work is available for Hierarchical S5 all-sky GW search #6
12/24/2009 10:14:52 AM Einstein@Home Message from server: No work available for the applications you have selected. Please check your settings on the web site.
Which is admittedly better then "client to old" but still no work...
But we mean it in the nicest possible way! ;-) Notoriety is just the price you pay for providing that magic piece of information that everyone now refers to. Of course, in my case the info just supported what I've always thought about Dell... (the laptop was given to me so I guess I really shouldn't complain much)
Mike