OK, we have a disagreement on whether modern servers should be considered descendants of minicomputers or microcomputers. As the terms are varied anyway, there is no "exactly right" answer, but the position seems to be while the software (UNIX, and NT with its obvious inheritance of UNIX and VMS design characteristics), and the hardware might have inherited some of the I/O tools of minicomputers, architecturally and financially they owe more to the micros of the 80's.
- Modern servers almost universally run some variant of UNIX or NT. Mini heritage there, but NT was put out by the quintessential microcomputer software company.
- Modern servers have single-chip microprocessors, either descendants of the x86 (and thus definitely with a microcomputer heritage), or a shrinking pool of RISC chips. Soon, it appears that we'll be left with just Intel x86-compatible chips (including the upcoming AMD Hammer 64-bit exteension) and the new Itanium architecture. Both come out of the microcomputer heritage. The original minicomputers had multi-chip CPUs - the single-chip CPU was architectural innovation that made the micro possible.
- x86-based servers have a motherboard design that evolved from the IBM PC.
- Modern servers are built by either: descendants of micro companies (Compaq etc.), descendts of "super micro" workstations (Sun), or IBM.
Would you also agree the term "minicomputer" is dead, and largely used to refer to historical PDP's, VAXes and the like.
Agreed that the term is now obsolete (I've been working with HP3000s for so long I think they'll never be obsolete, but HP is now calling them servers, so.)
Remember that there are a LOT of servers running non-Intel archetecture. Many of these are invisible (AS400s for example, as well as most HP UX servers; don't know what chips Sun is using) but they're still there doing their job. And many Intel-based servers are multi-chip (or at least multi-processor, if that's what you meant, as opposed to monolithic CPU). So I question above statement 2.
Finally with the last statment you dismiss the fact that IBM was the original mainframe company; and ignore HP which was really mainly a minicomputer company (but those darned printers sold so well!). Or maybe they were a calculator company. Depends on POV I guess.
I did like the cleanup whoever did on the opening statements tho. -justfred
I agree that the term is historical. The mid-range position is not, though.
You make very good points about the heritage of PC's from micros. I would counter with a few opinions:
- Yes, NT was put out by the quintessential microcomputer software company. It happened as that company hired the quintessential minicomputer OS architect, Dave Cutler, away from DEC. Allegedly he brought along DEC code (the proposed VMS successor Mica) and Microsoft later settled with DEC by paying them $150 million. See History of Microsoft Windows.
- The single-chip microprocessor was definitely the innovation that made the microcomputer possible. But other than being on a single silicon die, today's CPU's have far more internal architecture in common with minicomputer CPU's than they have in common with the 8086. Today's Pentium IV is more like a superscalar pipelined RISC CPU with an x86 instruction set emulation layer than it is like the 8086. Itanium's design is based more on PA-RISC than x86, with emulation layers for both. Alpha is, well, Alpha.
- And, of course, IBM was the mainframe and later mini (s34, 36, 38 before AS400, and don't forget RS6000) company.
I'm thinking the argument of more or less like one or the other is perhaps not useful, and I made a to-do over nothing.
Perhaps we should say something along the lines of "current personal computers evolved out of microcomputers by integrating the features of minicomputers" or something like that. I'll stick that in and you can revise as you see fit, in good Wiki style.
Thanks for the good debate --Alan Millar
Alan, right back atcha. (Don't know if I had a bad tone, I'm not trying to sound confrontational about this; tho I guess to some people it is more like a religious argument!) --Justfred