Apple Silicon M1: Black Magic Fuckery • Hacker News Board

Re: Apple Silicon M1: Black Magic Fuckery

#481

Post by seabrookmx » Wed, Nov 25, 2020, 6:45 AM UTC

thinbeige wrote: post #415

koffiezet wrote: post #188

I understand the machine is great or going to be great for most use cases. My mbp is my main workhorse, but as a freelance SRE "devops" guy, the Apple ARM platform won't be suitable for my job any time soon, if ever. Docker is not yet available - but even when it would become available, emulating virtualised x86 code is explicitly not going to be supported. That in many cases means pulling a docker image built in a c…

FWIW, I never run docker on my local machine (I develop on a remote machine), benefits: remote machine os + setup is very close to production and GBit bandwidth up and down at my hoster is so much nicer when working with Docker images.

VS Code Remote SSH makes this a really nice experience too.

Re: Apple Silicon M1: Black Magic Fuckery

#482

Post by ur-whale » Wed, Nov 25, 2020, 6:46 AM UTC

PSA: however impressive the M1 hardware is, you're still going to be stuck using OSX, playing in Apple's walled garden and being subjected to their awful policies.

I'll gladly join the groupie crowd once Linux runs stable on it.

Re: Apple Silicon M1: Black Magic Fuckery

#483

Post by tachyonbeam » Wed, Nov 25, 2020, 6:46 AM UTC

_delirium wrote: post #99

liamcardenas wrote: post #91

I know very little, so perhaps someone could enlighten me. But I am curious how Apple Silicon will be for machine learning. When Apple releases a MacBook Pro with 64GB of unified memory (assuming they will) — won’t that be amazing for machine learning? I am under the impression that GPU memory is a huge factor in performance. Also, is there any way that the neural engine can accelerate training — or is it just for ex…

I wouldn't expect it to be particularly competitive in training large models. It's an integrated GPU with 8 cores, and the "neural engine" has an additional 16 cores. The kinds of discrete GPUs (mostly Nvidia) that people use for deep learning have more like 5000+ cores. I think Apple is aiming more at either training small models, or running pre-trained models. For example Photoshop is starting to integrate neural f…

I think it's pretty clear they're aiming at inference only. Training models on laptops is never going to be competitive. Might be fun for prototyping small models in PyTorch/TensorFlow though.

Re: Apple Silicon M1: Black Magic Fuckery

#484

Post by choicenotchance » Wed, Nov 25, 2020, 6:47 AM UTC

Sadly it can't run docker, and as a primarily backend dev, it's a huge problem for me as docker is a very important part of my workflow.

Re: Apple Silicon M1: Black Magic Fuckery

#485

Post by CyberRabbi » Wed, Nov 25, 2020, 6:48 AM UTC

These ppl are so drunk on the kool aid they don’t even realize that 80% of their experience is common to every new laptop purchase. In 1 year these “revolutionary” computers will be boring and slow again, especially when their batteries start to deteriorate.

Re: Apple Silicon M1: Black Magic Fuckery

#486

Post by tachyonbeam » Wed, Nov 25, 2020, 6:49 AM UTC

m15i wrote: post #433

liamcardenas wrote: post #91

I know very little, so perhaps someone could enlighten me. But I am curious how Apple Silicon will be for machine learning. When Apple releases a MacBook Pro with 64GB of unified memory (assuming they will) — won’t that be amazing for machine learning? I am under the impression that GPU memory is a huge factor in performance. Also, is there any way that the neural engine can accelerate training — or is it just for ex…

I too am curious how 64 GB unified memory performs for training deep learning models. Even if speed isn't amazing, 64 GB is much greater than the 24 GB available in Nvidia's flagship consumer cards, which would allow for inputting larger images, bigger batch sizes, deeper networks etc. Also, will be interesting to see how all of the different cores are used.

It's only got 8GB or 16GB of RAM AFAIK, because the RAM is part of the chip, and more memory would mean a really big die.

Re: Apple Silicon M1: Black Magic Fuckery

#487

Post by tgma » Wed, Nov 25, 2020, 6:50 AM UTC

nottorp wrote: post #115

nojito wrote: post #101

Earlier quoted context omitted.

The quote right after explains your concerns. >The memory bandwidth on the new Macs is impressive. Benchmarks peg it at around 60GB/sec–about 3x faster than a 16” MBP. Since the M1 CPU only has 16GB of RAM, it can replace the entire contents of RAM 4 times every second. Think about that…

Sounds like some form of GDDR instead of plain DDR. Not only faster, but I bet simultaneously accessible from both the CPU and GPU.

Nope. Nothing too magical, I suspect. It says LPDDR4 right there in System Profiler, although it does not indicate the frequency.

Re: Apple Silicon M1: Black Magic Fuckery

#488

Post by girvo » Wed, Nov 25, 2020, 6:51 AM UTC

brundolf wrote: post #25

This is fascinating: > Retain and release are tiny actions that almost all software, on all Apple platforms, does all the time. ….. The Apple Silicon system architecture is designed to make these operations as fast as possible. It’s not so much that Intel’s x86 architecture is a bad fit for Apple’s software frameworks, as that Apple Silicon is designed to be a bespoke fit for it …. retaining and releasing NSObjects i…

Taking that as true for a moment, I wonder what other programming languages get a benefit from Apple's silicon then? PHP et al. use reference counting too, do they get a free win, or is there something particular about Obj-C and Swift?

Re: Apple Silicon M1: Black Magic Fuckery

#489

Post by EugeneOZ » Wed, Nov 25, 2020, 6:53 AM UTC

ur-whale wrote: post #482

PSA: however impressive the M1 hardware is, you're still going to be stuck using OSX, playing in Apple's walled garden and being subjected to their awful policies. I'll gladly join the groupie crowd once Linux runs stable on it.

Same here. I took my decision to leave macOS after their release of BigSur. It's absolutely, 101% unacceptable to me when someone else controls what program I can run on my hardware, and I should hope that their server will not go down.

I still congratulate the team of Apple Silicone. Hope it will force Intel to create something similar.

Re: Apple Silicon M1: Black Magic Fuckery

#490

Post by ogre_codes » Wed, Nov 25, 2020, 6:53 AM UTC

bitwize wrote: post #459

ogre_codes wrote: post #442

Earlier quoted context omitted.

I suspect the previous poster was referring to non-Apple ARM hardware, not Apple servers. > The wider point being that Apple will probably never let you run Linux on their hardware Apple has already issued docs on how to load alternative OSs on their system and has said explicitly that Windows support is up to Microsoft. Linux on Mac metal is not out of the question, but it's going to take some time to get running we…

That was for x86 Macs. Apple is not supporting alternative operating systems on the M1 Mac except via virtualization.

From the sounds of it, Apple hasn't put anything in the way of installing Linux or Windows on the M1 based Macs. They don't have the sort of built in support the Intel based Macs have, but they should be able to boot Linux or Windows on bare metal.

Recently there was a Apple support article posted here on HN detailing it.

https://news.ycombinator.com/item?id=25145119