Video Enhance AI - Crashing after windows10 update 2004

I’ve already been in contact with the support, i first though these crashes were related to me upgrading to a new version of Video Enhance but it appears to be related to my “feature update” of windows 10 from 1909 to 2004.

I cannot for the life of me render content using the Video Enhance software anymore, before the windows 2004 upgrade i could render for weeks on end without crashing, now it crashes randomly after 1-15 hours time, usually happens when i use the computer while rendering, it’s more rare to happen when i am not using the computer for anything else but it still happens.

I’ve now tried literally everything, i’ve even downclocked my nvidia 2080ti card to ridiculous slow speeds in order to test if it is the card that is suddently failing but nothing helps.

The error log reads:
2020-09-18 21:44:14.793208: E tensorflow/stream_executor/cuda/] could not synchronize on CUDA context: CUDA_ERROR_ILLEGAL_ADDRESS: an illegal memory access was encountered ::
2020-09-18 21:44:14.793214: E tensorflow/stream_executor/cuda/] Error polling for event status: failed to query event: CUDA_ERROR_ILLEGAL_ADDRESS: an illegal memory access was encountered
2020-09-18 21:44:14.793400: F tensorflow/core/common_runtime/gpu/] Unexpected Event status: 1

ANYONE have an idea how to resolve this? I’ve tried every nvidia studio driver available, i think i’ve managed to render 4 videos in a whole weeks time and failing around 25 of them having to start over, the worst one was today, 15 hours of rendering and 10 minutes left and it suddenly crashed, 15 hours of 500watts power usage just down the drain for nothing, absolute shit.

Yep, same here since latest windows update. I cant even preview or render anything, get a pop up box saying to update graphics drivers etc. then just crashes. Windows 10 Pro. Version 20H2. Build 19042.572

Couldn’t edit last post. Update: I went into preferences and changed options form GPU to CPU, program now doesn’t crash, this maybe help for those who have issues until it can be fixed.