PERSEUS-9, The Dual-6502 Portable Machine That Should Have Been

A question: does anyone who was around in the early days of the 8-bit computer revolution remember a dual-CPU 6502 portable machine like this one? Or just a dual-CPU machine? Or even just a reasonably portable computer? We don’t, but that begs a further question: if [Mitsuru Yamada] can build such a machine today with parts that were available in the era, why weren’t these a thing back then?

We’re not sure we have an answer to that question, but it just may be that nobody thought of it. Or, if they did, the idea of putting two expensive CPUs into a single machine was perhaps too exorbitant to take seriously. Regardless, the homemade mobile is another in a growing line of beautifully crafted machines in the PERSEUS line, all of which have a wonderfully similar look and feel.

For the PERSEUS-9, [Yamada-san] chose a weatherproof aluminum enclosure with just the right form-factor for a mobile computer, as well as a sturdy industrial look. Under the hood, there are two gorgeous wire-wrap boards, one of which is home to the 48-key keyboard and the 40×7 alphanumeric LED matrix display, while the other is a densely packed work of art holding the two 6502s and a host of other DIPs.

The machine is a combination of his PERSEUS-8 computer, his 6802 serial terminal, and the CI-2 floating point interpreter he built for the PERSEUS-8. A brief video of the assembly of this delightful machine is below. One of the many things about these builds that impress us is the precision with which the case is machined, apparently all by hand. How he managed to drill out all those holes for the keyboard without having one even slightly out of alignment without the aid of CNC is beyond us.

53 thoughts on “PERSEUS-9, The Dual-6502 Portable Machine That Should Have Been

    1. As a SuperPET owner of over 40 years (!) it still has a very special place in my heart. And on my bench as I still try to get to the custom SD card module design and build. Yes, I know they exist, but I have my own ideas…

      As a computer engineering student I spent a lot of time with both processors and with many of the “WAT” languages it supports.

  1. The Superbtain was a twin Z80 machine, and several second processors could be attached to the BBC microcomputer, including a second 6502 and a several Z80 extensions..

    1. Several late 70s computers were effectively dual CPU as used, since they had one in the terminal half. That’s why I guess that rolling into the 80s machines that managed it all on one chip seemed good…. since then you didn’t need to have a “computer” before you could have a computer.

      1. Polish K-202 (1970-1973) was (tho did it made to working prototype is another question) also multi cpu microcomputer architecture, tho original CPU was still made out of TTL chips.

  2. “if [Mitsuru Yamada] can build such a machine today with parts that were available in the era, why weren’t these a thing back then?”

    Expense, for one.

    But very impressive build once again from [Mitsuru Yamada].

    1. There were machine like that, with some degree of portatility and integrated non-CRT display, like the Olivetti P6066.
      For reasons like that was designed to do word processing and print invoices had integrated a printer inside.

      Was really expensive, more tah the Olivetti M24 – that was expensive too.

      For dual-CPU personal computer there were some designs too: the C128 had both a z80 and a 6510 out of the box, for instance.In Olivetti made a version of the M20 that had both a Z8001 and 8086 CPU. Wasn’t 100%fully IBM compatible.

    1. my first real experience with an apple ][ was on a machine with a z80 softcard and a controller for 8″ floppies. used it primarily as a cp/m machine. one of the engineers built a graphics card from a upd7220 for it and we wrote a driver for a cp/m fortran plotting package.

  3. I have some memory of a Z80 board in the form of an XT or AT expansion card and I think it could run it’s own OS independent of the PC. I guess there were similar cards with other CPU’s back then too.

    1. There were quite a few different ones. Like the z80 on a card for running CP/M… then there were terminal emulator or uplink cards that had the micro version of the mainframe CPU on the card, IBM did a couple of flavors of those. Then also it was semi-popular as dev boards for various new micros. Then later we got things like sega megadrive on a card.

  4. The Commodore C128 had both a 6502 and a Z80. It could boot into 3 mode, C64, C128 with extended memory/features/80 char display, and CP/M mode that also had 80 char display. I still have one. I rarely used it in any mode but C64.

    1. And if you had a C128D it had another 6502 in the floppy drive… and at least the plastic version came with a carry handle, so it was somewhat portable if we ignore the CRT.

      And then there was the truly portable SX-64, which thanks to the integrated floppy drive was also dual CPU.

  5. I love the display. Using that many HDSP intelligent LED matrix displays is B1-bomber-level expensive though. As a consumer product, maybe you’d have a single line of them, but even that would be expensive, hence why most “pocket” computers used an LCD.

      1. An 8 digit 5×7 led display has 280 die to attach and 280 wirebonds. That was quite a cost in it’s own right , ignoring the yield issue of having so many die. Wirebonding is expensive enough that we got a special variant of a 44pin micro made with only the 16 pins we needed bonded out. (We also saved by eliminating most of the wafer testing, and all the package testing, and letting our end of line process do testing, and remapping around bad memory, and even bad peripherals.)

  6. There was at least one article in Byte about runing two 6502s together.

    There was a 6502 handheld computer, I forget details, but it was on the cover of Byte

    1. Seems like there were several machines that took advantage of the 6502 only needing the bus for half a cycle by having the video system do its DMA in the other half. Stands to reason you could just slip a 2nd 6502 into that off cycle just by feeding it an inverted clock. Could use a dual port RAM if video were needed. Seems like the stack could be a problem, though.

      This sounds like fun. Quickly, Ginger, to the laboratory!

      1. The concept was there, but I can’t think of examples. I suddenly thought of a Japanese computer with two 6809s. A search says the Fujitsu FM-8. But one is used as the CPU, the other for graphics, and I don’t see an explanation of how closely they were tied together

  7. A Sinclair Spectrum was reasonably portable. You could easily put it in your school bag and plug it into your friends TV at their house.
    I am not saying that this is the same thing, just that this meant that there wasn’t so much of an incentive to make portables.

    The Grundy NewBrain (my Uncle had one) had a built-in VFD display, so that was usable as a true portable.
    The Wikipedia article mentions that a battery powered version was planned.
    Only a single Z80, though.

    1. I had à mate that had one of these – it had (for the time) pretty good graphics on the video output, and in particular a very mad flood fill algorithm – it looked totally random but ended up filling any space…

  8. Awesome craftsmanship!

    “Or, if they did, the idea of putting two expensive CPUs into a single machine was perhaps too exorbitant to take seriously.”

    Well, while isn’t exactly the same architecture as the PERSEUS, the Commodore 1541 disk drive was built itself around a 6502, similar to the one used in the C64. In fact, some demos (for instance were said to be able to offload 3D graphics calculations to the “secondary” CPU.

  9. i have vague memories of an article in (i think) dr. dobbs. the idea was that, since a 6502 only does stuff on the rising edge of the clock, you could build a dual-processor 6502 machine by inverting the clock to the second processor. as the clock rises, one processor does its stuff. as it falls, the other does.

    they are, however, only vague memories.

    1. I thought it was Byte. But I also saw similar schematics for the Z80 and the 6809. The 6809 one was in a Motorola application note.

      And yes, these were two processors running on an inverted clock. I’m still not sure what you got out of that, we had the hardware but nothing further. That’s what I was thinking of when I saw the title.

      Most of the examples thrown out here are a choice of CPUs. So OSI had a three CPU computer, but only one ran at a time. During the transition to the 8088, Godbout and the like sold boards with an 8088 and an 8085, but you’d choose between them. The Apple Z80 card was to run CPM, the 6502 was used too, but only for I/O. Gimix offered a serial and a parallel port that each had its own CPU. I have a 512K print spooler with a Z80, same concept, but not integrated into the main CPU.

  10. Game consoles had dual CPU. I think Sega Genesis/Mega Drive were the first major system with dual CPU: 68000 and Z80. Atari Jaguar had 5 CPUs and can be assigned to specific tasks, although some games were wasted on 68000 as the main CPU.

  11. The British BBC Micro computer can be installed with a 2nd 6502 as a co-processor. The 1st CPU handles all the I/O and display etc, the 2nd CPU does all the number crunching. Very expensive options back in those days.

    1. yep – we had a classroom full of BBC-B’s, and the server had the 2nd CPU and a dual 5.25″ floppy disk unit! Such power…. :) Admittedly it was pretty baller for the time

  12. 6502 expensive??? 6502 was designed as a low cost alternative to the MC6800. Something like $35.00 vs $100 when it was introduced, hence being at the heart of apple 2, Atari 800 et al.

  13. I’ve had good success drilling consistently-spaced holes with a drill press fence and a piece of protoboard. Drill the first hole, drill out a hole in the protoboard to be the same size, and bolt it down. Then drill successive holes in the protoboard and run them all the way through the metal behind it. As long as spacing is in multiples of 0.1 inches, you’re golden.

    1. Thanks for the great information. I am the author of this PERSEUS-9 project and was having a very hard time because I always drilled the holes next to each other by eye. I will try that method next time.

  14. Ummm did anyone else catch that the SW was hand assembled, entered via switches, and debugged… Then put into EPROMS with a home brew programmer? That’s old school – much respect for the effort. That development path was certainly chosen for the enjoyment of the effort – I applaud the craftsmanship.

    1. I am the author of this PERSEUS project. I thought I would try to recreate the efforts of people in the past, when PCs did not yet exist in the world as they do today and practical interpreters and compilers were not bundled or released to the public. I found that sometimes it is easier to understand the basics, from simple bit manipulation to floating point arithmetic operations. It is surprisingly fun to develop on an 8-bit CPU where we can make all the hardware and software ourself.

  15. Xerox had a dual CPU model in their 820 line. It had a Z-80 and an 8086, could run both CP/M amd CP/M-86 simultaneously. The 820-II could be upgraded with the 8086 board but not with the 3rd model’s larger CRT.

  16. Twin Z80s in the form of the Philips P2000C – A luggable that was surprisingly versatile. You could even add an 8088 CoPower card for MSDOS goodness and a total of three processors. One of my favorite “nobody has ever heard of it” machines. I wrote a lot of code for it back in the day.

  17. The Heath H89 / Zenith Z90 was another series of computers with dual CPUs. It was a Z80-based terminal that handled the screen and keyboard, connected to an internal Z80-based CP/M computer. The design split tasks nicely, resulting in a computer that was about twice as fast as a one-Z80 computer (like the Osborne or Kaypro).

  18. Even the earliest Commodore dual drives had 2 6502s and shared access to RAM built into the drive. With PET and that fast IEEE-488 interface, you have a 3 processor system- if only somebody made use of them.

    1. Motorola had two app notes on 68B09 dual-processor systems. I don’t know if the designs made it into any commercial systems, but they are certainly interesting. The weak point: although both use B-series 2 MHz parts, to get the timing right they ran at 1 MHz. I’m not sure for what applications it makes sense to buy two 2 MHz CPUs and run them both at 1 MHz, instead of buying one 2 MHz CPU and running it at full speed. There might have been small advantages for interupt handling, but that’s about it. Really the problem was that the DRAM of the day wasn’t fast enough to feed two 2-MHz 6809s.

  19. My Z-100 had an 8085 and 8088 CPU that would run concurrently. In fact, the CP/M BIOS was actually written in 8086 code, and left a (for the time) enormous 56K TPA for the 8085.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.