4 ms·
For reference, this is the original context: The title-linked reddit thread, in turn links to the following YT video, which is of the WWDC23 Apple Keynote spee
by crtified 3y ago
For reference, this is the original context:
The title-linked reddit thread, in turn links to the following YT video, which is of the WWDC23 Apple Keynote speech of June 5. See below the link for a text transcript of that relevant excerpt.
https://www.youtube.com/watch?v=GYkq9Rgoj8E&t=666s https://www.youtube.com/watch?v=GYkq9Rgoj8E&t=666s
"Finally, the 32-core Neural Engine is 40% faster [than M1 Ultra]. And M2 Ultra can support an enormous 192GB of unified memory, which is 50% more than M1 Ultra - enabling it to do things other chips just can't do. For example, in a single system, it can train massive ML workloads, like large transformer models, that the most powerful discrete GPU can't even process, because it runs out of memory."