توی لینک زیر، میتونید عملکرد GPU-ها برای اجرای LLM-ها رو باهم مقایسه کنید:
https://www.runpod.io/compare/h100sxm-vs-4090
به درد زمانی میخوره که بخواید GPU اجاره کنید. بنچمارک هم توسط runpod ارائه شده که خودش GPU اجاره میده.
BY PyTorch Howsam
A project of our size needs at least a few hundred million dollars per year to keep going,” Mr. Durov wrote in his public channel on Telegram late last year. “While doing that, we will remain independent and stay true to our values, redefining how a tech company should operate.
That growth environment will include rising inflation and interest rates. Those upward shifts naturally accompany healthy growth periods as the demand for resources, products and services rise. Importantly, the Federal Reserve has laid out the rationale for not interfering with that natural growth transition.It's not exactly a fad, but there is a widespread willingness to pay up for a growth story. Classic fundamental analysis takes a back seat. Even negative earnings are ignored. In fact, positive earnings seem to be a limiting measure, producing the question, "Is that all you've got?" The preference is a vision of untold riches when the exciting story plays out as expected.
PyTorch Howsam from tr