No joy! Using /dev/urandom's idea of two instances I got the Memory Alloc errors to go away by using the CPU on the elevated instance and miners on normal permissions. After using DDU and installing the 16,12,2 drivers I tried running both CPU/GPU as one elevated instance, running both as a regular instance, running an instance for each with both instances elevated, and running the CPU elevated with GPU normal. Every option with 3rd party workers crashes within 15 minutes.

Any other thoughts or ideas?