From: psychonaut@nothingisreal.com   
      
   Greetings.   
      
   On 2024-11-28 13:05, Carl Fink wrote:   
   > I have a system with a very high-end CPU (for a workstation), a 12th   
   > generation Intel i9, but a low-end GPU (Radeon RX470 with 6 gb of RAM).   
   >   
   > Is it possible to do any sort of AI training on that setup, say if you're   
   > willing to process at 5% of an nVidia GPU's speed? I just want to play with   
   > it, see if it's worth investing more money, at this point. I don't mind, if,   
   > say, training a LORA takes an hour per photo, or even 6 hours per photo.   
      
   Yes, it's absolutely possible to train generative AI on a CPU rather   
   than a GPU, but whether this is feasible depends entirely on what sort   
   of model you are expecting to end up with, how much training time you're   
   willing to tolerate, and how much RAM you have available.  Without   
   further details on your requirements it's difficult to make any specific   
   recommendations.  I've seen a few reports comparing CPU and GPU training   
   of LoRA models for Stable Diffusion, for example, that indicate that CPU   
   training can require four times more memory and/or forty times more   
   time.  If you don't already have the requisite GPU hardware and don't   
   want to buy it yourself, it might be economical to rent the processing   
   power from a cloud service.   
      
   Regards,   
   Tristan   
      
   --   
   =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-   
                      Tristan Miller   
   Free Software developer, ferret herder, logologist   
                 https://logological.org/   
   =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-   
      
   --- SoupGate-DOS v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)