Ruin-Capable@alien.topBtoLocalLLaMA•Speed and energy use - RAM vs Mac Studio vs RTX cardsEnglish
1·
1 year agoWhat model are you going to run that can accept 100GB of context?
What model are you going to run that can accept 100GB of context?
I wonder if you could get it running on two Mac Studio Ultras with 192GB of RAM each. With fewer nodes you’d reduce the communication overhead quite a bit.
Ooh… now I’ve got another model to play with. :D