10

Was the reason because he knew the 68K from the Macintosh, or were there other reasons that spoke in favor of the 68030 and the additional 68882 floating-point coprocessor? The i386, along with a 387 co-processor, would have had all the functions that were necessary for a then-modern operating system like NextStep. Thanks to its VM86 mode, the 386 CPU would also have allowed DOS programs to be run natively inside the VM86. That's why I'm wondering why the 386 wasn't chosen. ​ I don't know whether an i386 was cheaper than the 68030.

  • The Intel 80386 with 25 MHz was introduced in April 4, 1988 and had a performance of about 7.5 MIPS. 80386 with lower clock rates where available since 1985.
  • The Motorola 68030 with 25 MHz was introduced in around 1987 and had a performance of about 5 MIPS.
10
  • 2
    Comments have been moved to chat; please do not continue the discussion here. Before posting a comment below this one, please review the purposes of comments. Comments that do not request clarification or suggest improvements usually belong as an answer, on Retrocomputing Meta, or in Retrocomputing Chat. Comments continuing discussion may be removed.
    – Chenmunka
    Dec 12, 2023 at 18:44
  • What does 'performance of about xxx MIPS' actually mean? Can you add instead real performance measurements with some contemporary benchmarking tools like dhrystone?
    – lvd
    Dec 13, 2023 at 11:21
  • 4
    The custom DMAC, good for graphics blitting and other large data copying ops, was an important part of the NeXT hardware. Would the messy memory organization of Intel have gotten in the way of this?
    – John Doty
    Dec 13, 2023 at 17:24
  • @JohnDoty (or anyone): is memory organisation at all messy in 32-bit mode?
    – Tommy
    Dec 15, 2023 at 0:22
  • @Tommy In 32-bit mode, the user's virtual machine sees a simple memory layout, but this is a consequence of MMU machinations. The DMAC would have to reproduce those.
    – John Doty
    Dec 15, 2023 at 13:07

3 Answers 3

16

Very likely, familiarity with the Motorola environment and Motorola as a supplier within the development staff might have been technical reasons for choosing that CPU. After all, quite some of the designers had an Apple background, just like Jobs himself. But I don't think basing NeXT on Intel would have presented a major technical problem for any of their developers above maybe minor inconveniences (Keep in mind NeXTSTEP for Intel became available 1993).

Beyond technical reasoning, I would assume marketing reasons were considered much more important.

Looking at the first NeXT presentations, you can easily see that the Next ecosystem had the intent to appear as something revolutionary, new and ground-breaking, throwing all of the "old and awkward stuff" out of the window. The NeXTSTEP environment was demonstrated (quite credibly, at the time) as a revolutionary object-oriented and entirely user-friendly (and, admittedly, "exotic" and "luxury" - to justify the price point) combination of hardware and software.

Considering these ambitions, the added capability to be able to execute "boring old DOS programs" like WordPerfect or Multiplan would rather have de-valued the platform than been seen as added value, and, more importantly, would probably have encouraged reviewers to directly compare NeXTStations to high-end PCs, both price- and performance-wise, this totally defeating the ground-breaking targets (and, obviously, also eroding their $$-value). After all, DOS programs would in no way have profited from NeXTSTEP. Next wanted to be clearly distinct from the PC market, not integrate with it. So, the ecosystem targeted not at competing with high-end PCs, but rather at markets above.

I think it must be noted that this actually seemed possible at the time NeXT was active. The high-end DOS/PC platform was by far not as prevalent and "standard" as it was in later years, thus that marketing policy was probably even reasonable (and resonated quite well in the higher-education and academic research market, for example, and, as such, worked at least for some years).

5
  • 1
    At that time Unix vendors were still a thing and the target audience universities and number crunchers. Neither of these were particularly interested in DOS-compatability Dec 14, 2023 at 13:31
  • Another technical reason might be that Mach was already ported to m68k, whereas I am not sure about x86. Dec 15, 2023 at 7:50
  • 1
    @JörgWMittag Mach 2.0 (the base of what went into the NeXT) was available as ports for both m68k (Sun 3) and X386 (PC) platform.
    – tofro
    Dec 15, 2023 at 12:19
  • 6
    They had a bit of a weird relationship with the academic research market. My lab at MIT used NeXTs as lab bench machines: the DSP port was a great way to import data from an instrument, and concocting a GUI to operate one was not terribly difficult. NeXT noticed we were a big customer, so they sent a few people to visit and see what we were up to. We took them into the lab: typical physics lab with messy benches, whirring vacuum pumps, hissing nitrogen, great fun. There, on a bench, was a NeXT cabled to a vacuum chamber where we were calibrating an x-ray spectrometer (continued).
    – John Doty
    Dec 15, 2023 at 13:23
  • 6
    They were horrified! The acted as if we were torturing kittens. I think they expected that we were something like the Media Lab, doing artistic things, rather than merely making tools to explore the Universe ツ
    – John Doty
    Dec 15, 2023 at 13:27
19

As far as I can tell almost nobody except the PC makers used x86. 68k was the more common architecture by far being in the Sharp 68k, Amiga, Macintosh, Megadrive, Sun-1, Vaxstation 100, and SGI Iris. 68k was actually a very popular chip for workstations and that's what Next was making.

Making a workstation from the 386 would actually have been pretty strange.

11
  • Were there really 68k machines called Vaxstations? Dec 14, 2023 at 21:03
  • According to wikipedia at least the 100 was I guess the others may have just been vaxes and terminals now that I read it.
    – davolfman
    Dec 14, 2023 at 22:09
  • 3
    I think the Atari ST line was also 680x0.
    – Geo...
    Dec 14, 2023 at 23:12
  • 2
    @Geo... Yes, Atari ST was 68000, as were Sun-1 to Sun-3 systems. Dec 14, 2023 at 23:46
  • 1
    @ThorbjørnRavnAndersen - SGI used 68k, in the IRIS (not IRIX, which is an OS) machines, up to 1986/87 and then switched to MIPS. Dec 15, 2023 at 23:49
9

According to https://en.m.wikipedia.org/wiki/NeXT several people came with Jobs from Apple, including at least one hardware designer which presumably knew the 68000 family very well already.

My guess would be that it would be much faster to get a finished product out by continuing what they already had been doing and knew worked, as opposed to having to switch architecture to something unknown to them.

4
  • and they probably had code that they could reuse too Dec 14, 2023 at 11:51
  • 1
    @Jean-FrançoisFabre if I recall correctly the basis of the operating system was BSD with a Mach kernel and gcc as the compiler. Rest was most likely written from scratch. Next step was very different from macOS Dec 14, 2023 at 12:41
  • @Jean-FrançoisFabre But wasn't most of the code for Next not already written in high-level languages like C and Objective-C?
    – Coder
    Dec 15, 2023 at 1:21
  • I believe Apple still based their sw on Pascal. Objective C was if i understand the wiki and remembers correctly not used at all in the Apple MacOS environment (i used ThinkPascal in university, something which in many cases was replaced with Think C and later Think C++) Dec 22, 2023 at 18:53

You must log in to answer this question.

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