Several GPU workunits in a row failed, at least two dozen from various BOINC projects.

MilkyWay@Home v1.02 (opencl_nvidia):
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=1559971920
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=1559971926
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=1559971928

Milkyway@Home Separation (Modified Fit) v1.36 (opencl_nvidia_101):
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=1559971742
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=1559971744
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=1559971749
milkyway.cs.rpi.edu/milkyway/result.php?resultid=1559971751
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=1559971753

Part of a large cluster of GPU workunits from various BOINC projects that failed about the same time, so at least one of them did not leave BOINC and the driver in a state where another GPU workunit would start properly.

The OpenCL section of the Nvidia 364.72 driver, and earlier 364.* drivers, has a problem which can an entire computer to lock up, or cause a few dozen OpenCL tasks (often not all from the same BOINC project) to give a quick Compute Error. Problem not seen in the 362.00 driver.

Tasks from POEM@home seem the most likely to trigger this problem.

Threads on the problems:

https://www.primegrid.com/forum_thread.php?id=6769#94223

http://boinc.fzk.de/poem/forum_thread.php?id=1205#10896

The 365.10 driver is now available, but it does not fix this problem.