Since the 3060s would have to communicate constantly using pcie, I think I'd go ... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

Almondsetat 29 days ago | parent | context | favorite | on: Qwen 3.8 27B

Since the 3060s would have to communicate constantly using pcie, I think I'd go with intel

apitman 29 days ago [–]

Currently software is king when it comes to inference performance though.

Almondsetat 29 days ago | [–]

Having the entire model in the vram without having to transfer anything is even more king, though

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact