Faceswap Forum

Hi
I have been using Faceswap for the last year and had a lot of fun with it. However, some months ago after an update I couldn't no longer train Realface models. At first I believed that the update had made the system to require more 'system power', but then I saw that my GPU was not used any more
Based on the different support post that I have found in this forum I can see that the lines in the system output file:
gpu_cudu: No global version found. Check Conda package for Cuda
gpu_cudnn: No global version found. Check Conda packages for Conda cuDNN
maybee could explain my problem?

I have tried to:
1) delete/reinstall Faceswap
2) Uninstall MiniConda and delete/reinstall Faceswap
3) Uninstall MiniConda, clean registry with CCleaner and delete/reinstall Faceswap
4) Update the drivers for my 'NVIDIA GeForce GTX 1660 Ti' graphics card
5) Tried to install Cuda from the Nvidia home page
6) Am in progress of trying to get Cudnn in the same way...? But I am not sure this is the right way to go...?

Do some of you have a suggestion of what I should try?
This "Check Conda packages for Conda cuDNN". I do not have any idea of how to do that? Anywhere where I can learn further on that?

Several month ago an update of Faceswap made my Faceswap application crash every time I tried to run it. I followed some guidance in this forum that guided me to role back something. Can recall what it was, tensorflow or...? It was some thirdparty thing that was not working well and you were waiting for an update on that. The guide was great and I could run Faceswap again (I have not been able to find that guidance again?) I am wondering if it could be some leftovers from that, some registry setting etc. that blocks some part of the Faceswap installation?

Thanks and regards

DearForrest wrote: ↑Mon Jan 10, 2022 6:59 pm

gpu_cudu: No global version found. Check Conda package for Cuda
gpu_cudnn: No global version found. Check Conda packages for Conda cuDNN

No, this is correct, you don't want a global version installed unless you know what you are doing.

Do some of you have a suggestion of what I should try?
This "Check Conda packages for Conda cuDNN". I do not have any idea of how to do that? Anywhere where I can learn further on that?

Yes, see here and provide the requested info:
app.php/rules#rule-4b

Hi
While I was waiting to have my question published, I tried to make a clean Windows and install on that.
I started with the GPU for Windows. Then when trying to start Faceswap the console window was shown for 2-3 sec. and then closed without any further happening. I presume that some pre-requisite was missing, but didn't knew how to troubleshoot that, so I deleted the Faceswap folder and tried to install the CPU version. That could run.
Then I tried again to delete the folder and install the GPU version. Then this could run, but still without the power of the GPU. So I think I am back to my original problem

I have attached the system info, hope that can help

Thanks and regards,

gpu_cuda:            11.6

Uninstall your global CUDA. It is conflicting with the faceswap version :/

Will just say that I uninstalled all Cuda 11.6 things, then deleted the FaceSwap folder and unstalled miniConda. After that ran the Windows GPU installer.
It worked

Still in the task manager the GPU workload is as low as 1-3 %. However, the training goes on again with descent speed, just like it did long time ago before I ran into the trouble. So I think every thing now works as it should and am very happy

My word is final