8

Magic, which is described as a "smart" layout system for integrated circuits, was open-source and available by the end of 1983. As far as hardware and OS requirements, early documentation just says:

Magic is written in C under the Berkeley 4.2 Unix operating system for VAX processors.

Since it appears that Magic was quite early, successful, open-source, and even is still around 38 years later, I am wondering what hardware it was ported to and typically run on in its early years.

What particular machines ran Magic, and what hardware specs were needed for it to be usable circa 1985?

Anecdotal: I do recall using EDA software for PCB design and SPICE on an unaccelerated Amiga at that time, and finding it to be adequate. But based on the description of Magic, I'd be surprised if such an "advanced" VLSI layout tool was usable on a 68000. Perhaps it was run on higher spec Sun/HP/NeXT workstations, and possibly accelerated Amiga or Mac II?

4
  • An unspecified "VAX" is usually an 11/780, from when there was only one. That's nominally a 1 MIPS machine. The next couple of VAXen were smaller (11/730, 11/750). The 11/785 (1.5 times the speed of an 11/780) was not released until 1984.
    – dave
    Commented Dec 2, 2021 at 14:43
  • It’s even a 1 VUP machine ;-). Commented Dec 2, 2021 at 15:42
  • Not being familiar with the layout system, the title had me thinking the question is about Magic: The Gathering. To say the least, it was a little confusing… Commented Dec 2, 2021 at 16:30
  • I thought it was about the pre-emptive multitasking OS for the Atari ST/TT/Falcon. Commented Dec 4, 2021 at 21:56

1 Answer 1

6

The published version of the introductory paper says that

Magic is written in C under the Berkeley 4.2 Unix operating system for VAX processors and Sun workstations. The current VAX implementation works only with AED color displays with special Berkeley microcode extensions.

A Sun workstation back then would have been at best a 68010-based Sun-2. The VAX system referred to in the paper was a VAX-11/780.

The introduction to Magic 6, released in 1990, mentions that

Magic runs on a number of workstations, such as the DECstation 3100 and Sun's SPARC processors. Partial Unix System V support is provided, via the compilation flags mentioned below. The system also runs on the MacII.

Support for the Mac II was new:

To assist people with small machines (such as the Mac II), Magic can now be compiled without some of its fancy features. Compilation flags are provided, as indicated below, to eliminate things like routing, plotting, or calma output.

The build configuration explicitly mentions a number of supported hardware platforms: VAXen, MIPS (in little-endian and big-endian mode), WRL Titans, m68k systems (Sun-2, Sun-3, Macintosh II, HP 9000 series 200/800/820), SPARC, and IBM PC-RT. The build supports BSD-based systems (such as Ultrix), SunOS 4.0 or later (using SunView or X11), A/UX and HPUX.

Since support for smaller systems was new in 1990, I think it’s fair to imagine that only “larger” systems were supported in 1985. (“Larger” taken with a pinch of salt, since a Macintosh II in 1990 would be about as capable as a 1985 Sun-2; but by 1990 the main Magic platform had moved to the MIPS-based DECstation 3100 and SPARC-based SPARCstation 1.) Given the conference publication dates of the Magic papers, in 1984 and 1985, there’s even a decent chance that Magic hadn’t grown much beyond its original systems by 1985.

It’s also possible that the limitations on the Macintosh II are tied to limitations in the hardware platform rather than its performance; PMMU-based Macs running early versions of A/UX (before version 3.0) didn’t support demand paging for example.

I haven’t found details of Magic releases between the system described in the papers and Magic 6, apart from a couple of mentions of the PC version of Magic 4.

4
  • I wonder if the Mac II port was more precisely an A/UX port. If A/UX was popular anywhere, I'd guess late-80s California academia.
    – Brian H
    Commented Dec 2, 2021 at 14:58
  • Yes, it was an A/UX port (which is why the limitations of A/UX are probably relevant, see my second-last paragraph). Commented Dec 2, 2021 at 15:39
  • Ah, I see. So Magic may have required a pretty large amount of memory (real or virtual) for full functionality. And maybe that's what is implied by "larger systems".
    – Brian H
    Commented Dec 2, 2021 at 15:42
  • Yes, that’s what I’m thinking too, although the program does go to some lengths to limit its use of memory, and a 1985 machine wouldn’t have had all that much anyway. If memory really was a big problem, I would have expected some mention somewhere, e.g. “you can use this on a Mac II but you need x amount of RAM”; but I haven’t found anything like that. Perhaps it seemed obvious at the time... Commented Dec 2, 2021 at 15:45

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .