I have just installed 6.6.36 and my Cuda will not run.
it stays in "Waiting to run"

I have checked Preferences and it is set to run when comp and Gpu in use, and I have suspended & resumed the Wu, the project and restarted the machine.

I run Win XP Sp3 32 bit

Do I need to reset the long term debt?

Did you also check local preferences, in case you ever set those (Advanced view->Advanced->Preferences, processor usage), that your GPU is set there as well to go do work? Try a Clear anyway.

No specialties in a cc_config.xml file?
No separate venue set up that that computer is using, so you changed the preferences in the wrong venue?

---

Duh.....I looked at "use Gpu while computer in use" and I saw it was unchecked.

It must default to off at install..........

Thanks, Problem fixed


I looked at it but I did not "see" it

lol, I have those days as well. :-)
Nothing to be ashamed of, we're only human.

---

6/17/2009 8:59:17 PM rosetta@home Requesting new tasks for GPU


and other projects.....


this showed up on my log...... rosseta has no gpu app....

v. 6.6.36

i've downgraded back to 6.31 and it's correct now., not getting request work for gpu from a project that doesn't have gpu app..

---

I recommend Secunia PSI: http://secunia.com/vulnerability_scanning/personal/

I'm gettting this message all the time:

2009-06-18 10:15:22 Milkyway@home Sending scheduler request: To fetch work.
2009-06-18 10:15:22 Milkyway@home Requesting new tasks for GPU
2009-06-18 10:15:27 Milkyway@home Scheduler request completed: got 0 new tasks
2009-06-18 10:15:27 Milkyway@home Message from server: No work sent
2009-06-18 10:15:27 Milkyway@home Message from server: CPU app exists for MilkyWay@Home but no CPU work requested

.. for various projects that doesn't have GPU work. Most of them don't care, but some does, and this hinders me from getting work.

Is a downgrade the only fix for now? Is the BOINC software team aware of this problem?

rgds
Peter

i've downgraded back to 6.31 and it's correct now., not getting request work for gpu from a project that doesn't have gpu app..

v6.6.31 (and in fact the entire v6.6.xx range) will ask for GPU work 'periodically'. The very earliest ones would, after an initial flurry, settle down to asking once per day: later ones will ask after random time intervals, averaging twice per day.

The only relevant change between .31 and .36 is that BOINC is now, at long last, explaining a little bit more about what it's doing. So the only thing you have gained by downgrading is ignorance ;-)

The BOINC team do know that these GPU requests are being sent - they wrote the code, after all! I think the idea is that if/when a project launches a new CUDA application, BOINC will notice and start using it without you having to make any changes by hand.

Then, if I understand you right, the client will also ask for CPU work, so this is really no problem?

One tend to notice only the problems and not the success messages :-)

Then, if I understand you right, the client will also ask for CPU work, so this is really no problem?

One tend to notice only the problems and not the success messages :-)

Yes, exactly. That's the big difference of v6.6.xx above v6.4.x: it separates the CPU and GPU much more completely. Sometimes it will ask for GPU work, sometimes it will ask for CPU work, and sometimes (rarely) it will ask for both together.

You probably noticed the GPU messages because they will come quite frequently at first (in case there is a CUDA application, but the server has just gone to sleep), but it will quickly calm down to the frequency I suggested previously.