ChipTuner on Nostr: That math doesn't math what engine are you using? A 256k context window needs like ...

That math doesn't math what engine are you using? A 256k context window needs like 100gb or more of working memory (vram or shared) + the model