Instant Nuke crash on creating nodes?

Hello everyone - I just installed Eddy_for_Nuke-2.7.2-nuke12.2-win64 and the very first video tutorial on the site asks me to load an E-Element node and Nuke instantly crashes, every time. Is this happening to anyone else?

Eddy version: EddyNuke 2.7.2
Nuke version: 12.2v2
Operating system: Windows 10
GPU model: GeForce RTX 2080 Ti
Graphics driver version: 461.72

Here is the Nuke output window after the crash:
NukeX 12.2v2, 64 bit, built Aug 4 2020.
Copyright © 2020 The Foundry Visionmongers Ltd. All Rights Reserved.
Licence expires on: 2021/9/14
Disk cache C:/Users/DANTER~1/AppData/Local/Temp/nuke\ViewerCache/??: 240MB (1% of 102400MB) used in 5 files.
EddyNuke[INFO ] - EddyNuke 2.7.2 (b5b8bc33bc603fad97ef8c2dff7253a7e42d6d2c)
EddyNuke[INFO ] - Copyright© 2019 VortechsFX Ltd. All Rights Reserved
EddyNuke[INFO ] - Portions of Eddy are subject to a perpetual license from Weta Digital Ltd.
Eddy[INFO] - EddyEngine 2.7.2 (55e6f183024e739879880aef98a0f052ec8a342a)
Eddy[INFO] - Copyright© 2019 VortechsFX Ltd. All Rights Reserved
Eddy[INFO] - Portions of Eddy are subject to a perpetual license from Weta Digital Ltd.
Eddy[INFO] - CUDA driver version: 11.2
Eddy[INFO] - CUDA runtime version: 10.0
Eddy[INFO] - Automatically selecting best CUDA devices.
Eddy[INFO] - Active GPU Devices:
Eddy[INFO] - Using GPU Device 0: “GeForce RTX 2080 Ti”, compute capability 7.5, 11264 MB
Eddy[INFO] - Inactive GPU Devices:
Eddy[INFO] - Inactive GPU Device 1: “GeForce RTX 2080 Ti”, compute capability 7.5, 11264 MB
Eddy[INFO] - Inactive GPU Device 2: “GeForce RTX 2080 Ti”, compute capability 7.5, 11264 MB
Eddy[INFO] - Inactive GPU Device 3: “GeForce RTX 2080 Ti”, compute capability 7.5, 11264 MB
Eddy[INFO] - Inactive GPU Device 4: “GeForce RTX 2080 Ti”, compute capability 7.5, 11264 MB
Eddy[INFO] - Inactive GPU Device 5: “GeForce RTX 2080 Ti”, compute capability 7.5, 11264 MB
Eddy[INFO] -

Thanks in advance for any help!

Additional thoughts- after reading some bugs that individuals were having by their connection to the onboard graphics port vs directly in the video card:

I’m running six local nvidia RTX 2080tis. I realize Eddy will only see/use one of them. Two are connected via the motherboard PCIe slots and 4 are connected to a breakout box via another PCIe slot. In the above Nuke output window, Eddy activates only one (as expected), but I wonder which one. Perhaps it grabbed one on the cards in the breakout box? The two monitors however are directly connected to one video card. I could try turning off the four other cards but Windows has a fit trying to bring all six back online. Before I pull the plug, any thoughts or obvious things I’m missing?

Thanks!

Hi,

You can try using the EDDY_DEVICE_LIST environment variable to change which GPU will be picked by Eddy. Set it to the number of the GPU to use, the output from Eddy should then show it using that GPU.

Eddy usually will work with whatever GPU you choose, it doesn’t have to be the one connected to the monitor, although we haven’t tested with a configuration like yours. You can run the command “nvidia-smi” from a command prompt to see what numbered GPU is connected to the display. Also “nvidia-smi -q” will display a lot of information, might be useful.

Hi Niall -
Thanks for the assistance. I changed the environment variable to each GPU and none of them did the trick of keeping Nuke from crashing upon adding an E_Element node.

I tried to load an example Eddy as well (there are four in the folder, as opposed to the dozen or so shown in the manual). When I try to load eddy eddy_combustion_sparse_example_01, I get a new error that states:
EddyNuke[ERROR ] - EDDY API error code=1 “Input plug ‘mesh’ has already been added to the layout”

Would you have any other suggestions to get the plugin to function?

Thanks,
Dante

Hi Dante,

There should be 19 example scenes included with Eddy. If you only see 4 then it sounds like something might have gone wrong with your installation. It will definitely cause issues and errors if other files are missing too.

There is a known bug in 2.7.2, although it hasn’t caused this error for other people yet. Try opening the eddy_quickstart_smoke_example_01 instead and see if you still get an error. Also if you can send us the whole output when it crashes while loading a file, it’s possible that might help.

Thanks,
Niall.