Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
boesboes
on Oct 12, 2023
|
parent
|
context
|
favorite
| on:
M2 Ultra can run 128 streams of Llama 2 7B in para...
Unless you calculate power usage I’d bet
datameta
on Oct 12, 2023
[–]
I would wager the M2 is more energy efficient than a 4090.
smoldesu
on Oct 12, 2023
|
parent
[–]
It fully depends on the workload. The 4090 itself draws around 450w at load, and the M2 Ultra peaks around 300w. If your workload is >1.5x faster on Nvidia hardware, then it's per-prompt efficiency probably beats out the M2 Ultra.
frontierkodiak
on Oct 12, 2023
|
root
|
parent
[–]
Not to mention that you can trivially set a 300w power cap on the 4090 and still get 75-80% of peak FLOPS.
datameta
on Oct 12, 2023
|
root
|
parent
[–]
All valid points, I think I looked at it too trivially and fired an opinion from the hip.
Consider applying for YC's Spring batch! Applications are open till Feb 11.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: