The First 'Apple Silicon' – The Aquarius Processor Project | Hacker News Reader

The First 'Apple Silicon' – The Aquarius Processor Project

thechipletter.substack.com

66 pointstim_sw17 comments

That was the first Cray I ever saw. I don't think I knew anyone who was a chip designer back then (at apple or anywhere else) so that shows how leaky the "security" was.

When I actually programmed one (a few years later, at NASA Ames) it was clear that running Unix (UNICOS) on a Cray was stupid. The machine was designed for highly pipelined vector FORTRAN and the continual branching of the C model was fog slow. Really it should have been designed as a vector processor fed by a host rather than as a standalone machine.

The author of the article paints a picture of a situation in which expensive hardware was purchased on a whim without much planning beforehand. This hardware was ill suited to their task, and ultimately led to failure of the project while others (e.g., ARM) were quickly producing RISC designs with low budgets.

This sort of thing isn't uncommon in larger organizations that have money to burn on R&D. As a result, it's probably safe to say Apple was in a position that allowed them to burn a lot of money in the late 1980s.

> In due course, the team found that the Sun workstations that they attached to the supercomputer could run their software just as quickly as the Cray itself.

I’m going to give the technical management the benefit of the doubt here, and assume the $15M Cray wasn’t primarily for chip design, but rather architectural simulation.

This was intended to be a parallel system across several layers, while wrapping in a number of novel architectural ideas. It would not have been ridiculous to assume having a massively parallel supercomputer to experiment with those ideas more quickly would be worth the expense.

The Cray was put to good use doing plastic injection molding simulations that saved millions of dollars on prototype mold tools. This was the era when the desktop Mac folks were making insanely complex screwless plastic cases that basically snapped together.

Oh I remember those cases! They seemed brilliant… until snap parts got stressed and snapped themselves, or worse when they “got you” snapping back in.

Seymour Cray was reported to have used a Mac for part of the Cray Supercomputer design work.

It was a joke: when Seymour was informed that Apple was using a Cray machine to design their computers he said something like "funny, I use an Apple computer to design the Cray". But in reality he never used computers at all but did everything with paper and pencil.

IIRC, for similar reasons, American Express purchased a Cray to AI fraud detection. I guess it was a giants rules engine. An former Amex exec reported to me it was a waste of money and the cheaper statisical models worked much better.

isn't this the hobbit/crisp core that was supposed to go in the newton?

No, Hobbit was a descendent of the AT&T Bellmac processor.

The end fate of Hobbit at Apple is not unrelated to ARM either (https://tompittard.com/media) :

> "Tom Pittard and Paul Gavarini of ATG spotted the ARM architecture (late 1985), rightly inferring Acorn had shared 6502 experience and motives. Investigation began. Pittard and Gavarini went to work on the Möbius project. Their two-year effort created a board using a VLSI Technology ARM2 chip plus software emulating the entire Apple desktop computer lineup: a 6502 or 65C816 running Apple II code, and a 68000 running Mac code. All emulated code ran faster on the ARM2 processor than the corresponding native versions. ... Möbius terminated quietly - but Pittard kept the prototype in his office." ... "In the afterglow of Möbius, Tom Pittard did a presentation for Apple engineering teams on his benchmarks a few months after work halted. Heads turned. Of particular note were the results from Pittard's Lisp benchmarks on ARM2." ... "Tesler [Lary Tesler], a veteran of Xerox PARC and an expert on GUIs, saw Pittard's results. He grasped where the ARM architecture could take Newton [Apples first handheld device]"... "Tesler had an endgame in mind... He based that on the status of Hobbit [a competing processor design], Tom Pittard's ARM2 benchmark results, and roadmap discussions with the ARM development team."

We hired one of the guys who worked on Apple's hobbit/crisp or the newton, he'd git his cpu back from the fab, worked first time, but politics suddenly went ARM - he quit

Also no, hobbit was a CRISP (RISC) core, bellmac aka WE32x00 was a CISCy (68020 level) core

I said it was a descendent, meaning it evolved from the Bellmac/WE32X00.

"The 20MHz 32-bit ATT92010 Hobbit, developed at AT&T’s Bell Labs, combines both complex and RISC design approaches. It can produce the highly compact code of a complex instruction set microrpocessor, critical in portable applications where memory space is at a premium, but like RISC designs, it supports single cycle instruction exectution and, hence, high performance"

Source: AT&T press release.

https://techmonitor.ai/technology/att_formally_launches_its_...

I spent a lot of time programming WE32X00s, they were very much similar to 68Ks/32ks - CISC descendants of vaxes. While I didn't get to program any of the CRISP cpus as I understand them they are much closer to 29Ks/SPARCs (RISC, with a form of register windows)

I did an assembly language course on the WE32100 and it was a pretty nice chip.

When I tell people today about the STRCPY opcode their eyes light up. First in amazement that the opcode existed, then in horror at the damage that it could potentially cause.

Reply on news.ycombinator.com