Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
tbruckner
on Aug 9, 2023
|
parent
|
context
|
favorite
| on:
Making AMD GPUs competitive for LLM inference
Isn't it closer to 80%? Where is it 90% of a 4090?
tails4e
on Aug 10, 2023
|
next
[–]
You're right, my mistake. Still cost and power per compute still favours AMD
meragrin_
on Aug 9, 2023
|
prev
[–]
Article specifically says 80%. My guess is mixing up the numbers between 4090 and 3090 TI.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: