1. X
  2. Adam Mainz
Log inSign up
Adam Mainz
4,160 posts
Adam Mainz profile banner
user avatar

Adam Mainz

@MainzOnX
PyTorch TPU @google | ex @meta | Obsessed with AI & ML Performance | Building what’s next 🔥 | Views are my own
github.com/adamomainz
Joined December 2023
323
Following
5,469
Followers
RepliesRepliesArticlesArticlesMediaMedia

Log in or sign up for X

See what’s happening and join the conversation

Continue with phone
or
Log in with username or email
Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
  • Pinned
    user avatar
    Adam Mainz
    @MainzOnX
    Jul 16
    Article cover image
    Article
    Two speeds of a GPU
    GPUs have two speeds A GPU has two speeds, not one. It has a speed for math (how fast it can add and multiply and generally push numbers around) and a separate speed for moving bytes between its...
  • user avatar
    Adam Mainz
    @MainzOnX
    1h
    Shabbat shalom ❤️
    Image
  • user avatar
    Adam Mainz
    @MainzOnX
    6h
    How you like your steak shows if you as a person are rare or if you are just cooked
    user avatar
    vlad
    @radmadvlad
    21h
    Medium rare beef vs well done is what separates the first world from the rest.
  • user avatar
    Adam Mainz
    @MainzOnX
    12h
    GO GO WATCH LIVE
    user avatar
    Yacine Mahdid
    @yacinelearning
    13h
    and we are live to talk about RLM as compositional generalizers x.com/i/broadcasts/1…
  • user avatar
    Adam Mainz
    @MainzOnX
    13h
    I will keep screaming about this topic until it’s in your brain the way it should. Memory and compute bound analysis is the easiest way to start prioritizing your performance wins. Don’t optimize the wrong things at the wrong time
    user avatar
    Adam Mainz
    @MainzOnX
    Jul 16
    Article cover image
    Article
    Two speeds of a GPU
    GPUs have two speeds A GPU has two speeds, not one. It has a speed for math (how fast it can add and multiply and generally push numbers around) and a separate speed for moving bytes between its...
Image
REPLAY
user avatar
Yacine Mahdid
@yacinelearning
RLMs are Compositional Generalizers with Alex Zhang from MIT
Advertisement
Advertisement