5 ms·
My biggest complaint is that I have a "16-core Neural Engine" in my MacBook, but nothing that can be run on it. Sure, internal macOS tools such as the fingerpri
by addandsubtract 4y ago
My biggest complaint is that I have a "16-core Neural Engine" in my MacBook, but nothing that can be run on it. Sure, internal macOS tools such as the fingerprint reader or even webcam might use it, but not a single ML project on github makes use of it. I can be lucky if ML projects don't depend on CUDA to run.
- Tepix 4y agoHere's one project https://machinelearning.apple.com/research/stable-diffusion-coreml-apple-silicon https://machinelearning.apple.com/research/stable-diffusion-... https://huggingface.co/blog/diffusers-coreml https://huggingface.co/blog/diffusers-coreml Not surprisingly, they are quite a bit slower than those power-hungry nVidia GPUs.
- thecupisblue 4y agoHonestly, I think I'll sell my MacBook Pro and just buy a new one with less cores. Bought it expecting to be able to abuse it towards ML, but it's barely even supported.