I seem to have a problem. The 2 tasks running on my duel core machine (an Hadam3p and a Hadcm3n) have paused. They now say waiting to run (waiting for GPU memory). What gives? Are the WUs trying to run on the GPU. This project does not support GPU processing. This is a laptop with integrated graphics. How do i correct this? Please help.

---

That's a new one! You're correct, this project has nothing to do with GPU-anything. I'm grasping at straws here: Have you tried old faithful, normal shutdown and reboot?

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

Also, you're running other projects. Are any of these for GPU work?
And do you have the graphics display open, or are you trying to open it?

Remember, with GPU work, this is using the display card/chip that normally is used for painting whatever goes onto the screen. So you can only do one of these at a time: display OR crunch GPU work.

If you're doing GPU work,then perhaps the chip, which uses ON BOARD MEMORY, is using so much of what is there, that there isn't enough left for also running cpu work, in this case 2 climate models.

---

Backups: Here

Rebooting seems to have worked. The problem has cleared and the WUs are running again. I am attached to 2 other projects (Milkyway@home and World Community Grids new Malaria drug search), but, did not have any of their WUs on the machine at the time.

---

Hi JIM, etal,

I had this problem a while back and posted for some help with it here...

(Waiting for GPU Memory) Status OSX:
http://boinc.berkeley.edu/dev/forum_thread.php?id=6966

...the suggested solution (at the time) was to upgrade to BOINC 6.12.41 (a beta version), I was (and am) at 6.12.35 on a Mac OSX 10.6.8. However, I was in the middle of several CPDN Tasks at the time and declined to go that route.

FWIW, my best guess as to the cause of this problem was that it seemed to precipitate whilst I was experimenting with both the online and BOINC Manager preferences and switching between the two in a haphazard fashion...I never could recreate the issue on demand, but it always occurred after such tamperings on my part...what it appeared to me to be was this, if an active WU was using both active and virtual memory and preferences were changed mid-stream, this would cause the WU to cause this hiccup...manual purging of the VM using Activity Monitor (OSX) was sufficient in resolving the issue, though, as you noted, a reboot would do the same...I've since settled on a set of preferences that I like and the problem has never resurfaced.

Hope that's helpful... :)

Thanks for the explanation, what I understand of it. I am using Win7 and the problem disappeared after a shut down and reboot. Fortunately, it has not recurred.

---

:)