Hardcore! 350-page report reveals why Apple M1 is so powerful (with PDF download)

The M1 chip released by Apple last year shocked the technology world with its incredible IPC and extremely high energy efficiency, and for the first time, people's attention was truly shifted from the x86 architecture to the de facto high-performance architecture.

However, despite the powerful performance of the M1, the ecosystem of this chip still adheres to Apple's usual closed style. Anyone who wants to use the M1 hardware must go through Apple's own operating system Mac OS, otherwise it will be very difficult.

Despite the difficulties, many engineers have begun working to overcome them, and recently someone even managed to run Linux completely on the M1 chip.

Also continuing this attempt is Maynard Handley, one of the former developers of Apple Quick Time. Recently, he shared a 350-page PDF analyzing the internal workings of the M1 ARM chip.

This document (currently version 0.70) delves into the M1 architecture from a reverse engineering perspective and has been greatly supported by other professionals and enthusiasts in the field.

The amount of work involved in this analysis report cannot be underestimated, as it requires a deep understanding of the ARM architecture and its diagnostics, as well as hours of testing, technical documentation and patent literature reading, and community discussions.

This document is a collection of known, existing information and reverse engineering research results on the M1 chip, and also includes the author's own experimental results.

While the report is not yet final, Handley's efforts can now become a community-driven M1-geared bible. Parts of the report can also lay the foundation for further decryption of Apple chips in the future, such as reverse engineering the more powerful Apple M2.

Hopefully, this report will open the door for closed system proprietary hardware platforms to move onto more popular open source platforms.

Maynard Handley released the document on Twitter.

Download address: (need ladder)

https://drive.google.com/file/d/1WrMYCZMnhsGP4o3H33ioAUKL_bjuJSPt/view

The report was originally written for myself and requires some hard-core knowledge.

The author said that this article was written for himself, but the audience is anyone who is interested in the technical details of the M1 chip. Of course, readers need a relatively high level of basic CPU knowledge.

Let’s take a look at his own introduction to the report:

I'm pretty confident about everything I discussed in the previous section on L1 cache, after I had time to run experiments, read patents carefully, and edit my writing.

The discussion of L1 cache, and everything that follows, is more tentative. It's possible that this section was written without any experiments (or that I wasn't happy with some of the experiments, because I was always thinking of new ways to interpret the results, so the results were inconclusive).

From the many patent documents I have referred to, I think the basic idea of ​​this report is correct, but this conclusion is the result of a quick glance and analysis, and I have not thoroughly read or tracked all relevant patent documents.

That said, the A15 and its companion chip (perhaps the M2) will be available soon and people will want to start looking into it. Hopefully this document can serve as a primer so that later people can spend more time on fundamental analysis to really understand new things, rather than spending time reinventing the wheel.

This document was actually written for myself to record my investigation and research into the M1 chip. These investigations took the form of experiments, with reference to many Apple patent documents, and all the reported contents were linked together by a reasonable knowledge of the academic literature.

Anyone interested in the technical details of the M1 chip is a potential reader of this report. This report assumes that the reader's knowledge of CPUs is much higher than that of the average Internet opinion leader. The report includes a large number of references to papers and patents. Reading patents and papers requires hard work.

I suggest that readers skip any parts that do not appeal to them, whether it is the way the experiments were designed, the way they were interpreted, the description of the literature surrounding a certain point, or the patent literature.

I have done my best to report accurately. However, there may be multiple errors in the report, whether in the experimental design, implementation, analysis, my understanding of the patent, or any other aspect. Technical corrections are welcome.

If you are confident in your technical skills and interested in Apple's patents, you can download it and take a look.

https://worldrepublicnews.com/learn-all-about-apples-m1-with-this-350-page-deep-dive/