Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

My M1 MBP is built and still performing so well, it’s going to last at least 3 more years.

Unless Apple integrates a new accelerated ML chip next year that’s going to revolutionize the way we’re able to train and interact with small, specialized AI models in macOS.



I’m really curious when we will get there. Especially game consoles, a dedicated chip for running inference could be really fun for developers


I thought Nvidia cards already had some kind of ML hardware for upscaling video games? Which seems them going full circle, since large AI models (except Google TPU code) are usually trained on GPUs, which were originally intended for games.


I was thinking that you’d need a separate chip so that there’s no memory issues when rendering, but I don’t have enough expertise in this


The Tensor cores on nVidia GPUs are effectively "a separate chip", they're just on die. That's where they can run DLSS and frame generation ML inference.

Separating them out would hamstring them, since they wouldn't be able to process the frames as they're being rendrered without performance penalty.


ANE saw an improvement from M1 to M2, but I’m hoping for a more significant one for M3




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: