Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
derision
on Dec 13, 2019
|
parent
|
context
|
favorite
| on:
Deep Learning Breakthrough Made by Rice University...
Latency between GPUs kills performance
sudosysgen
on Dec 13, 2019
[–]
It depends on just how huge the model is. Some models take multiple seconds to run/backpropagate and might take hundreds of gigabytes of memory, in which case it could be useful.
strbean
on Dec 14, 2019
|
parent
[–]
Also seems like a problem that could be partially solved by tailoring the NN architecture. Does that make sense?
ganzuul
on Dec 14, 2019
|
root
|
parent
[–]
Do you mean like Stochastic Gradient Descent does?
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
