- Joined
- Dec 19, 2005
- Messages
- 18,455
"Meta builds its own AI accelerators, and at Hot Chips 2026 it presents the next two: the MTIA 300 for training and the MTIA 400 for generative AI and recommendation. Recommendation compute at Meta grows 3x a year, and GPUs sit idle waiting on memory for those workloads. I also show the first two MTIA generations on camera, and you see Meta's new MX8 format, with one shared exponent across a whole matrix."
View: https://www.youtube.com/watch?v=Y7VigDmUds8
View: https://www.youtube.com/watch?v=Y7VigDmUds8