Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Apple unveils M3, M3 Pro, and M3 Max

apple.com

Post reply on HN
Print view View original

11–20 of 1001 posts

Re: Apple unveils M3, M3 Pro, and M3 Max

#11

Post by jdprgm »

I don't understand why they don't update the entire lineup whenever a new chip comes out, especially if really only swapping the chip in the macbook pro and imac here.

On a separate note does anyone have any insight on how unified memory compares to vram in the context of machine learning performance? Considering an H100 with 80GB costs like $30k a maxed out macbook pro with 128gb unified memory for $5k is interesting. Is it remotely comparable or compelling for large models considering realistically most prosumers are capped at the 24gb nvidia cards for anything resembling a reasonable budget.

Re: Apple unveils M3, M3 Pro, and M3 Max

#16

Post by throw0101a »

I'd prefer if there'd be higher RAM limits for each variant:

* ~~Ultra~~ Max probably has an adequate limit with "up to" 128GB

* Pro would be nicer to have "up to" (say) 64GB, rather than 'only' 36GB

* Base perhaps "up to" 32GB, rather than 24GB

There are some situations (running a VM or three with different OS(es)) where there really isn't a substitute for more bits.

Re: Apple unveils M3, M3 Pro, and M3 Max

#17

Post by mardifoufs »

minimaxir wrote: post #4

> Additionally, support for up to 128GB of memory unlocks workflows previously not possible on a laptop, such as AI developers working with even larger transformer models with billions of parameters. The presentation explicitly highlighted this very randomly, without showing how AI development actually works. I know Apple has put more effort into Apple Silicon support for PyTorch but is it there yet?

I'm pretty sure pytorch supports it OOTB. In my experience it works even better than ROCM
Post reply on HN