Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
jedberg
on Aug 27, 2024
|
parent
|
context
|
favorite
| on:
Cerebras Inference: AI at Instant Speed
Cheaper on a per token basis.
striking
on Aug 27, 2024
|
next
[–]
I accept that. Your original comment left me under the impression that this represented a shift closer to the edge (I still don't think the hardware is all that much smaller), but I'll agree this is cheaper per token under full utilization.
sanxiyn
on Aug 27, 2024
|
prev
[–]
Doubtful. SRAM is not cheap, and this is entirely about SRAM vs HBM.
jedberg
on Aug 27, 2024
|
parent
[–]
They list the price in this press release. So either they're taking a big loss or they're doing it cheaper per token.
throwup238
on Aug 27, 2024
|
root
|
parent
|
next
[–]
It wouldn't be the first time a manufacturer ignored capital amortization to post better numbers.
sanxiyn
on Aug 27, 2024
|
root
|
parent
|
prev
[–]
Not even necessarily a loss let alone big, since their comparison includes margin.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: