This just started happening on my two CUDA systems. One yesterday and then again today on the other system.

This WU got -12 error and from that point forward, all CUDA WUs get -9 overflow or -12 errors. I don't have access to logs at the moment, but scanning my tasks list this appears to be the earliest one that errored.

XP Pro-32b w/SP3, Driver is 181.20 (I believe...don't have access to system at the moment) AMD X2 4000 cpu.

Had my wife shut the system down as it's just dumping WUs like crazy for now.

My other CUDA system, after rebooting yesterday, seems to be running fine again. Cycling BOINC did NOT clear the problem as the gpu seemed to be fubar'd.

By the looks of it, there's something stuck in memory. I have forwarded it to the developers, nonetheless.

---

By the looks of it, there's something stuck in memory. I have forwarded it to the developers, nonetheless.

Thanks. As I said, only a reboot clear the loop, so the gpu has been screwed up somehow.

Additional info now that I'm home to collect it.

This WU got error messages:

05-Mar-2009 08:49:51 [SETI@home] Task 12ja09ae.28485.172479.3.8.180_0 exited with zero status but no 'finished' file
05-Mar-2009 08:49:51 [SETI@home] If this happens repeatedly you may need to reset the project.
05-Mar-2009 08:49:51 [SETI@home] Restarting task 12ja09ae.28485.172479.3.8.180_0 using setiathome_enhanced version 608
05-Mar-2009 08:50:17 [SETI@home] Computation for task 12ja09ae.28485.172479.3.8.180_0 finished

After the restart it finished with -9 overflow. All WUs that followed got -9 overflow or -12 error until system was shutdown.

I have restarted the system and there does not seem to be further problems with it.