http://www.gpugrid.net/workunit.php?wuid=5096726
For the above work unit, I shut down and then restarted my computer. When the system came back up, the screen kept blanking, and I got a system tray popup that said the 331.20 Nvidia driver stopped working, but successfully recovered. However, the system quickly became unresponsive, then blue screened and rebooted. It did this a couple of times. I eventually rebooted into safe mode, ran BOINC, suspended the task, then booted back into normal windows. As soon as I resumed the task, the driver problem started again - however, this time, I was able to suspend the task before the blue screen occurred. I decided to abort the WU. Another WU downloaded and there the same problem has not reoccurred.

Has anyone seen this before? If so, is there a way to recover without aborting the WU?

Thanks.

---

That was a SANTI task. Many Win7/8 users are having trouble with SANTI. Try the SANTI errors thread for suggestions.

---

BOINC <<--- credit whores, pedants, alien hunters

I suggest you add a Boinc start delay to a cc_config file, to allow you sufficient time to abort the troublesome WU's:

<cc_config>
 <options>
  <start_delay>30</start_delay>
 </options>
</cc_config>

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

I have seen that a few times on my 660's with a Santi WU. The only way to solve it was what you did: Windows in safe mode and then aborting the WU. I did not yet find any other way to get it solved.
Skgiven suggestion for the delay is very good, as you then have time to abort the WU, before a BSOD.

---

Greetings from TJ

Until we can get application version 8.15 installed on the long queue. I'm not quite sure what the delay is there.

Here is a thread that, I believe, is the same problem.
http://www.gpugrid.net/forum_thread.php?id=3491

As Richard said, GPUGrid application version 8.15 fixes the problem, but has not yet been deployed to the Long queue.
http://www.gpugrid.net/apps.php

Sad, really.

If it's a frequent issue, stick to the short queue, which uses 8.15 apps.
The work around can result in having more than 2 GPUGrid WU's per GPU. This might be the reason it was not applied towards the long queue. It might be research related (doubt it though) or just logistical (the only person that can do it didn't and isn't available).

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Supposedly 8.15 has been deployed to the long queue. So, feel free to get long queue tasks to see if the issue persists.