Hacker News Reader

Comment by Krish1577 | Hacker News Reader

Reverse-engineering Apple's custom silicon is basically the modern equivalent of repairing a spaceship while it's actively launching. Incredible work by the Asahi team!

View on HN

Makes you wonder who's choice it was to launch a half-finished rocket. AMD and Intel's rockets sit on the launch pad with GPU/CPU drivers shipped for Linux before T-10.

Their Silicon was written to be OS agnostic, Apple's was not.

>Their Silicon was written to be OS agnostic, Apple's was not.

i don't think that's accurate. the M series chips and memory architecture enable capabilities for the OS software to take advantage of, and the software and hardware teams get to collaborate to be ready for a simultaneously launch, but nothing about that is macos specific, just that they're on the same side of a walled garden.

Apple silicon has features specifically designed for macOS or software running on macOS. e.g.: Rosetta 2.

From Wikipedia:

One of the key reasons why Rosetta 2 provides such a high level of translation efficiency is the support of x86-64 memory ordering in the M1 SoC.[12] The SoC also has dedicated instructions for computing x86 flags.

So, Apple always adds one more thing to their hardware to enable their magic.

[0]: https://en.wikipedia.org/wiki/Rosetta_(software)#Rosetta_2

i wonder if this is not something that asahi could eventually make use of

edit: https://lwn.net/Articles/970907/

ya but that one is usable on linux, here is a real world project taking advantage of it: https://github.com/FEX-Emu/FEX/pull/2665

you can find more apple-specific stuff by digging around in these docs: https://asahilinux.org/docs/platform/subsystems/

The reason Apple has the in house ability to run on multiple chip systems is because of their history going all the way back to Next and that little handheld from the early 1990s under Scully and all their work with Motorola and IBM over the years, that ability is what separates them from Microsoft who is having trouble getting Windows working with multiple CPU chips. Apple is simply better at it.

Windows in the 90s was ported from x86 to MIPS, DEC Alpha, PowerPC, and SPARC.

Windows NT was originally developed for Intel i860XR, then ported to MIPS R3000. 386 was its third supported platform. Danox is just a hopeless Apple fanboy.

You're replying to an LLM-written comment.

Why do you believe that comment was LLM-generated?

The dubious metaphor is what tipped me off in the first place, but then I looked at the account activity. Dormant for over a year, submits vibecoded AI project to Show HN (which gets immediately downvoted and flagged due to being slop), then leaves two LLM-style comments.

macOS ships with full working drivers for everything. They're unfortunately not under AGPLv3.

They publish documentation because they need OEMs to be willing to support their products.

Apple doesn’t. Apple consumes it, so Apple always has a customer.

If selling Linux servers (or parts used in Linux servers) is a major chunk of your revenue, your company is likely to start providing first party Linux drivers?

This is a shocking development.

Apple's PCC standard supports Linux. It's just that Apple doesn't, so Apple Silicon plays second fiddle to Grace ARM cores and TPU hardware in the datacenter market.

Nothing prevents Apple Silicon from competing with Nvidia's ARM datacenter hardware, besides Apple. The only reason the rackmount Mac had to be depreciated is because Apple realized they were LARPing as a server hardware business.

Reply on news.ycombinator.com