Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
nullc
9 days ago
|
parent
|
context
|
favorite
| on:
AMD's Turin: 5th Gen EPYC Launched
They're memory bandwidth limited, you can basically just estimate the performance from the time it takes to read the entire model from ram for each token.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: