18

In the 1980s and 90s, there was a fad among the IT industry press to dub the newest "hot" microprocessor on the market as being a "mainframe-on-a-chip". I have seen this fawning description applied to the Mototola 68000 and 68030, the Intel 80386 and 80486, and the Zilog Z80000. Looking at those particular chips, there's definitely important technical innovation for single-chip microprocessors there.

  • 68000 Early 32-bit microprocessor
  • 68030 Advanced on-board MMU
  • 80386 Virtual 8086 mode hardware virtualization
  • 80486 Advanced on-board FPU
  • Z80000 Z80 fan's vision for world domination

But I don't see why any or all of these innovations would truly advance the objective of replacing an actual mainframe with a single-chip microprocessor system.

My question is was there ever a single chip microprocessor that fulfilled the role of directly replacing some aging mainframe system of its day? I think if the mainframe "platform" could be successfully replaced by using the microprocessor to run the same OS and applications (with at least source code compatibility), that should qualify. But I suspect that genuine mainframe applications probably had I/O performance requirements that greatly exceeded what the micros could do.

7
  • 6
    The first real "mainframe on a chip" was probably the Raspberry Pi emulating an IBM System/360 en.wikipedia.org/wiki/Hercules_(emulator) . At least it's somewhat chip-sized
    – tofro
    Dec 29, 2017 at 21:39
  • 68000 is a 16 bit processor.
    – JeremyP
    Jan 2, 2018 at 10:22
  • 5
    Mainframes were about handling a lot of diverse I/O jobs at one time (keeping them away from the actual CPU as much as possible), and later (S/370 etc.) generations were also about serious virtualization ("partitioning") capabilities (which came into the PC world later from the 80386 onwards - but not used at full capability until years later) - not raw, single threaded, general purpose MIPS ... Jan 2, 2018 at 14:09
  • 1
    @JeremyP, my sincere apologies - I couldn't see the words '32-bit' anywhere in the OPs question and thought yours was a belligerent line from nowhere. Then I saw '32-bit' - right next to '68000'! Sorry, me being daft :-) But it's fair to say that the 68000 chip implements a 32-bit register/addressing architecture, regardless of whether it uses a 16-bit ALU and other circuitry. The software platform it provides is 32-bit.
    – TonyM
    Jan 5, 2018 at 10:14
  • 1
    Maybe some microcontrollers come closer to it than computing-intense desktop CPUs. Mainframes were about I/O efficiency not numeric brawn - same applies to many MCUs. Aug 2, 2019 at 21:08

5 Answers 5

35

Yes, IBM System z mainframes (and their predecessors) have been using "mainframe-on-a-chip" microprocessors for a couple decades now. In 1995 I used a IBM PS/2 with an IBM System/390 Processor Card in it running MVS. It executed System/390 instructions natively, using (I believe) one of the same microprocessors used in the System/390 mainframes of the time.

According to Wikipedia:

Introduced in 1994, the six generations of the IBM 9672 machines were the first CMOS, microprocessor based systems intended for the high end. The initial generations were slower than the largest ES/9000 sold in parallel, but the fifth and sixth generations were the largest and most powerful ESA/390 machines built.

So in 1994 at least there were microprocessor-based mainframes capable of running real world applications.

Around the same time I also had a Unisys '486 PC under my desk with a Unisys SCAMP (Single Chip A-series Mainframe Processor) card in it, but I just used it as a PC to play games on after work. I don't know if the microprocessor used in this card was used in the regular Unisys A-series mainframes of the time.

7
  • 3
    @BrianH The one I used was just for development, but it probably could've been used for certain real workloads. I as I recall it had a fairly large RAID array, more than it would've needed just for development. I just used it as example of how small a "mainframe" could be. The same microprocessors (or ones like it) were used in full scale System/390 mainframes (about the size of a refrigerator or two) at the time.
    – user722
    Dec 29, 2017 at 23:28
  • 3
    @BrianH I found an article about using one of the systems I used in a production environment: "These P/390 systems have the power to make a departmental processing system. We have figured that one of these systems would be able to handle all of the POS processing for a large department store with a VTAM connection to a central site for doing file transfers. Or, it would be possible to run all of payroll on the system, and use NJE for sending the output (checks, ledgers, etc.) to a central site running the "large" printers." naspa.net/magazine/1996/May/T9605005.PDF
    – user722
    Dec 30, 2017 at 1:10
  • 2
    @RossRidge: Yeah, you could do that, but IBM really didn't want to sell the P/390 that way, because they wanted to sell you a smaller 390 mainframe or AS/400 for that job. In my experience, the P/390 was a dev/test/QA machine for the mainframe, and if you tried to do much else with it, IBM sales threw up all sorts of roadblocks (suddenly, no P/390s were available for 6 months! All discounts were being "reevaluated"! There just happens to be a giant incentive package on an AS/400!). Jan 3, 2018 at 7:39
  • 1
    @KJSeefried I think IBM may have been two minds here. The sales department wanted to sell more expensive models, but one document I found provided exampled configurations for a "VSE Small Application Server", a "VSE 'Old Iron Replace'", and an "OS/390 Large 'Data Mining' Application Server". ps-2.kev009.com/p390/$P390manuals/P390_qrefp390.pdf
    – user722
    Jan 3, 2018 at 15:14
  • 2
    @RossRidge Nice find...I don't think I ever saw that doco. Having worked for IBM, it wouldn't surprise me a bit that there were multiple sales groups pushing multiple, opposing, sales agendas. Ran into that all the time. However, if the P/390 sales team and any of the Large Systems guys crossed swords, there was generally only going to be one winner. :-) Jan 3, 2018 at 18:11
8

Let's assume we're talking "real" Mainframe instruction sets (and don't argue too much about what that means). Previous commentators mentioned the common ones:

  • IBM S/370 Based: XT/370, AT/370, P/370
  • IBM S/390 Based: S/390, P/390
  • Unisys A-Series: SCAMP Micro-A

There are a couple of more obscure ones:

  • NEC SX: The SX-ACE is a single chip vector/integer processor on a PCI-e card. Unfortunately, you can't buy it outside of one of NECs supers right now. But earlier you could get the NEC SX-6i and later SX-8i which were single-node (multi-chip module) deskside versions of the SX architecture.
  • NEC PX7x00: These are single-chip versions of the NEC ACOS mainframes, though they are always packaged as redundant pairs.
  • Fujitsu GS21: Single-chip descendants of the FACOM mainframe are still being sold.
8

Well, this question does lead deep into the gray area of what is a microprocessor and what has to be on chip to qualify as replacement, or do multi chip solutions also count, or when is a second chip an add on or an extension?

Not to mention the basic question what part of a mainframe is considered the CPU. Does the memory interface belong to the CPU or is it part of the memory (subsystem). Do IOCs count as part of the CPU or separate?

At least there is not doubt what a mainframe is, as there's nothing but /370(ish) (*1).

It maybe even does come down to the question, when is a microprocessor a microprocessor? There are not many chips that can be considered useful without supporting chips. A 8086 can only be used in very simple systems without 8282 and 8284 chips to decode its bus signals to be understood by 'normal' I/O or RAM/ROM.

Having said all that, there are many machines, that could be seen as microprocessor based /370 and alike:

Firstmost might be the IBM 5100. While not based on a single chip CPU (mind you, that was 1975), its discrete CPU did provide a mostly /370 instruction set.

Next there was the XT/370 in 1983, and its follow ups with the (main) implementation as one microprocessor. While the machine wasn't particularly fast (*2), it did fit with the lower end of what IBM offered as mainframes at that time.

But as you already mentioned, the true capability of a mainframe is its I/O performance. And that's something not even the later (1988) MCA implementation could do like the big ones.

During that time not only IBM, but also other mainframe manufacturers tried to sell microprocessor-based /370s. For example SIEMENS got its PC-2000, a single board add on for its Multi Bus based PC-MX2 Unix machines (NS32016 based). Fujitsu offered a similar setup. Again, due to the rather low I/O performance, they were only designated for development.

And then* there were also IBM (and other) mainframes using single chip CPUs, or better chip module, implementations starting in the late 1980s. First they were confined to the lower end, but nowadays next to all are based around such implementations. Strictly they are not single-chip but, then again, a Pentium II also is not.

Last but not least, today many lower to mid range systems are no longer based on 'real' implementations, but rather standard microprocessors running a /370 emulation. Fujitsu offers several mainframes based on Intel Xeon processors. Their actual machines can even be partitioned to run mainframe OS(es), Windows and Linux in parallel. This development is based on technologies they acquired when taking over Siemens-Nixdorf. Who again started it (codename Sunrise) when Siemens spun off their chip division (Infineon), so creating their chips couldn't be done in-house any more. At that time it was SPARC based, but now it's moved over to Intel. Still, the top end models are still based on custom implementations.

Before shunning emulation as 'not being the real thing' one should keep in mind, that almost every mainframe, beginning with the first /360s, was in part or complete delivered via a micro-programmed environment that changed many times below ISA level. The mainframe architecture never was about a certain real implementation, but it's an abstraction layer - or better a collection thereof.


*1 - Ok, there is another kind: Scientific processing with CDC and Cray as main players. But this kind never really had the pressure to put up with decades of ISA compatibility. Scientific applications, if they ever were used for more than a few years, had all source code available and just ported to the next generation anyway. There was never a real need to develop compatible microprocessor-based solutions - they got replaced right away by stock microprocessors.

*2 - 0.1 /370 MIPS may not sound fast, but at the same time a .9 MIPS machine with 1,25 MiB of memory was good to serve realtime data for 200+ individual terminal users.

2
  • @Raffzhan: "Scientific processing with CDC and Cray as main players. But this kind never realy had the pressure to put up with decaded of ISA compatibility." Counterpoint: NEC SX, a very successful, compatible ISA, scientific vector super first introduced in 1983 and still shipping. But yes, agreed, virtually everyone else went commodity + massively parallel. Even NEC went Xeon as I/O (host) processors on the latest SX (SX-Aurora) with the vector chips on PCI-e cards. Interestingly, the commodity world is sorta going that direction as well, with Xeon + Phi, x86_64 + GPU, etc. Jan 3, 2018 at 18:28
  • @KJSeefried agreed - with the liddle addition, that NEX's SX is only compatible to its own definition, unlike the /370 world where (code) cmpatibility reaches across generations and manufacturers.
    – Raffzahn
    Jan 3, 2018 at 18:29
6

Does VAX count?

MicroVAX 78032 was a microprocessor compatible with previous, multiple-chip CPUs and used in new VAXen.

7
  • 1
    Nop. Not a mainframe, just an overrated toy :))
    – Raffzahn
    Dec 29, 2017 at 19:33
  • 1
    Vaxen are considered to be minicomputers although the largest are arguably in the mainframe space at the bottom end.
    – JeremyP
    Jan 2, 2018 at 10:24
  • Minicomputer's the word ... Jan 2, 2018 at 13:57
  • @Raffzahn: Blasphemy. :-) Jan 3, 2018 at 7:25
  • 1
    On the other hand, the NVAX+ microprocessor (built in CMOS) was almost the power of the contemporary VAX 9000 "mainframe" (built in ECL), which systems sold for $1M US and up.
    – dave
    Mar 22, 2022 at 1:40
0

There have been a number of minicomputers-, mainframes-, and supercomputers-on-a-chip, including:

  • Chris Fenton's Homebrew Cray-1 on a Xilinx Spartan-3 1600 FPGA chip (Cray-1 supercomputer)
  • DEC J-11 (PDP-11 minicomputer)
  • DEC NVAX (VAX minicomputer/mainframe)
  • Harris HD6120 (PDP-8 minicomputer)

Please feel free to add to this list.

3
  • 1
    There's a little problem with this list: These are minis, not mainframes.
    – Raffzahn
    Mar 2, 2018 at 3:09
  • @Raffzahn The VAX 9000 is reportedly a mainframe. I've just added minicomputer and mainframe designations to the list, thanks! Mar 2, 2018 at 4:05
  • The DEC KL-10 was generally called a mainframe. This was the third processor in the PDP-10 family following the KA-10 and the KI-10. It ran TOPS-10 and later TOPS-20. It had a Motorola processor at the center, running microcode that emulated the PDP-10 instruction set. Feb 26, 2021 at 11:11

You must log in to answer this question.

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