Faceswap Forum

cloud7556

Like it says, to use the faster clustering with that many images you'd need more than 700gb of RAM. It uses exponential amounts of RAM to cluster based on how many images you have. You can split it to subses of 50k images to sort it in 4 packs if you want to use the faster method. thank you! I got ...

cloud7556

11/24/2021 16:53:03 WARNING Not enough RAM to perform linkage clustering. Using vector clustering. This will be significantly slower. Free RAM: 8373MB. Required for linkage method: 700884MB I tried to sort 200000 face images, but the system prompts that RAM is not enough. I know that if I wait all ...

cloud7556

https://forum.faceswap.dev/app.php/rules#rule-4b By manually installing tensorflow, I can now launch faceswap for extraction and training. But the speed is very slow, even slower than AMD GPU, and the CUDA utilization rate is 0%. I think I've done it When installing tensorflow manually, the CPU ver...

cloud7556

torzdf wrote: ↑Mon Nov 22, 2021 11:36 am
app.php/rules#rule-4b

By manually installing tensorflow, I can now launch faceswap for extraction and training.
But the speed is very slow, even slower than AMD GPU, and the CUDA utilization rate is 0%.

cloud7556

C:\Users\Administrator\faceswap>"C:\Users\Administrator\Miniconda3\scripts\activate.bat" && conda activate "faceswap" && python "C:\Users\Administrator\faceswap/faceswap.py" gui Setting Faceswap backend to NVIDIA 11/22/2021 19:23:49 INFO Log level set t...

cloud7556

torzdf wrote: ↑Thu Nov 18, 2021 11:48 am
Reinstalling Faceswap is unlikely to make a difference. However AMD really is not well optimizied for ML, so this may just be expected behavior (not ideal, I know, but Nvidia is much better at this stuff)

Thanks!
If I change to NVIDIA GPU, can I load the AMD training model for conversion?

cloud7556

RX580→RX6600XT
The speed even slowed down，Should I reinstall faceswap?

Faceswap Forum

Search found 7 matches

Re: Not enough RAM to perform linkage clustering. Using vector clustering. This will be significantly slower.

Not enough RAM to perform linkage clustering. Using vector clustering. This will be significantly slower.

Re: Error on launch: There was an error importing Tensorflow

Re: Error on launch: There was an error importing Tensorflow

Error on launch: There was an error importing Tensorflow

Re: The extract speed is not improved after upgrading the GPU?

The extract speed is not improved after upgrading the GPU?