this post was submitted on 18 Nov 2024
174 points (78.1% liked)

memes

10383 readers
2002 users here now

Community rules

1. Be civilNo trolling, bigotry or other insulting / annoying behaviour

2. No politicsThis is non-politics community. For political memes please go to [email protected]

3. No recent repostsCheck for reposts when posting a meme, you can only repost after 1 month

4. No botsNo bots without the express approval of the mods or the admins

5. No Spam/AdsNo advertisements or spam. This is an instance rule and the only way to live.

Sister communities

founded 1 year ago
MODERATORS
 
you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 2 points 4 days ago* (last edited 2 days ago)

The linear algebraic computations performed on their GPU's tensor cores (since the Turing era) combined with their CUDA and cuDNN software stack have the fastest performance in training deep neural network algorithms.

That may not last forever, but it's the best in terms of dollars per FLOPS an average DNN developer like myself has access to currently.