Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
ummonk
on Feb 23, 2024
|
parent
|
context
|
favorite
| on:
Stable Diffusion 3
Image diffusion models tend to have relatively low memory requirements compared to LLMs (and don’t benefit from batching), so having access to 128 GB of unified memory is kinda pointless.
Filligree
on Feb 23, 2024
[–]
They do benefit from batching; up to a 50% performance improvement, in my experience.
That might seem small compared to LLMs, but it isn't small in absolute terms.
ls612
on Feb 23, 2024
|
parent
[–]
I got a 2x jump on my 4090 from batching SDXL.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: