Question about 30xx support

Discussions about research, Faceswapping and things that don't fit in the other categories here.


Locked
User avatar
only5theboat
Posts: 3
Joined: Sat Oct 24, 2020 12:06 pm
Has thanked: 1 time

Question about 30xx support

Post by only5theboat »

I was looking for a more general thread to post this but couldn't find one. Any updates on support for 30XX series cards? I'm planning on doing a 5950X / RTX 3080 combination but awaiting the release of Zen 3 as well as the restocking of the 30XX cards.

User avatar
torzdf
Posts: 2649
Joined: Fri Jul 12, 2019 12:53 am
Answers: 159
Has thanked: 128 times
Been thanked: 623 times

Re: Question about deepfakes

Post by torzdf »

2 things need to happen for 30xx support:

1) Tensorflow needs to release a version with Cuda 11 support.
2) I will need to update Faceswap to latest TF api.

I can't do 2 until 1 is done, and I don't know how far off 1 is.

It will happen though.

My word is final

User avatar
only5theboat
Posts: 3
Joined: Sat Oct 24, 2020 12:06 pm
Has thanked: 1 time

Re: Question about 30xx support

Post by only5theboat »

One more question if you don't mind answering. I've been using DeepFaceL@b and it has alright success. Any key differences between faceswap and facelab? I will have a chance to experiment with faceswap tonight.

User avatar
abigflea
Posts: 182
Joined: Sat Feb 22, 2020 10:59 pm
Answers: 2
Has thanked: 20 times
Been thanked: 62 times

Re: Question about 30xx support

Post by abigflea »

There are many. I feel its faster due to the work on supporting the models with mixed precision .
The UI is way more intuitive.
The manual alignments tool is pretty sweet.
A bit more intuitive for sure.

:o I dunno what I'm doing :shock:
2X RTX 3090 : RTX 3080 : RTX: 2060 : 2x RTX 2080 Super : Ghetto 1060

User avatar
torzdf
Posts: 2649
Joined: Fri Jul 12, 2019 12:53 am
Answers: 159
Has thanked: 128 times
Been thanked: 623 times

Re: Question about 30xx support

Post by torzdf »

And the flip side of that is there is also very little difference ;)

The basic engine that drives both DFL and FS is the same. Just there are some features that FS has that DFL doesn't and some that DFL has that FS doesn't. You can get decent results with both.

I always say, try both. Use whichever works best for you.

My word is final

User avatar
JoeMoeDoe
Posts: 1
Joined: Wed Dec 23, 2020 5:46 pm

Re: Question about 30xx support

Post by JoeMoeDoe »

It seems step 1 is fullfilled:

According to this page
https://www.tensorflow.org/install/gpu

"CUDA® Toolkit —TensorFlow supports CUDA® 11 (TensorFlow >= 2.4.0)"

So now it would be good to work on Step 2 to get the RTX 30xx support in faceswap.

User avatar
bryanlyon
Site Admin
Posts: 793
Joined: Fri Jul 12, 2019 12:49 am
Answers: 44
Location: San Francisco
Has thanked: 4 times
Been thanked: 218 times
Contact:

Re: Question about 30xx support

Post by bryanlyon »

Locked