non CUDA devices - DirectML ? #2288
Replies: 4 comments 5 replies
-
|
I have the same problem, Have 6750 xt i dont know how use this proyect with DirectML, please help us !! |
Beta Was this translation helpful? Give feedback.
-
|
Same concerns, any support for non CUDA GPUS? or even just switching to CPU? |
Beta Was this translation helpful? Give feedback.
-
|
You should be able to use CPU just by installing a cpu version of torch and the TTS-cpu package |
Beta Was this translation helpful? Give feedback.
-
And waiting 8 hours for do a small process? Lool |
Beta Was this translation helpful? Give feedback.
-
You could try BARK, it have directml support or using ROCm in linux |
Beta Was this translation helpful? Give feedback.
-
|
TTS can run anywhere Pytorch can run. So it is more about PyTorch than TTS |
Beta Was this translation helpful? Give feedback.
-
|
xxx bark doesnt support amd at all its just a wrapped out of ancaoda which carry 30-40gb itself and the offical repo doesnt have amd support at all the person who made it make completely unable try urself pip install it fail else u need anacoda junk |
Beta Was this translation helpful? Give feedback.
-
i saw this if it helps |
Beta Was this translation helpful? Give feedback.
-
|
https://rocm.docs.amd.com/projects/install-on-windows/en/latest/how-to/install.html#hip-install-full check for compatibility if so you can run cuda via HIP DSK :) |
Beta Was this translation helpful? Give feedback.
-
Hi all
I did a search on this and only came up with one thread that only had one response that wasnt sure - BUT
Im wondering if there's any possibility of a DirectML port of this? Just for those who dont have CUDA gpus but have GPUs and hardware with ML capabilities (like the 7900xtx)
Would it be as simple as installing pytorch-directML and tensorflow-directML-plugin instead of its original?
Beta Was this translation helpful? Give feedback.
All reactions