Loading ...
Sorry, an error occurred while loading the content.

Re: Maybe a 1802 laptop ??

Expand Messages
  • Richard
    I have not given up on the idea, but the last 18 months have been really interesting and led directly to looking for a new employer recently and starting there
    Message 1 of 140 , Jan 7, 2013
    • 0 Attachment
      I have not given up on the idea, but the last 18 months have been really interesting and led directly to looking for a new employer recently and starting there next monday. Also, my old PC died back then, taking the hard disk and its contents with it.

      Since then I have only had little time for building computers, but I have at least a good concept:

      I have bought a bus card with 10 96 pin connectors, which should be sufficient. If I use Euro sized boards, it will fit into a Micro ATX PC case, wich is cheap and easy to get and also has an adequate power supply.

      I began with the memory map for the CPUs:

      0000 - 3FFF EPROM (16k):
      With more than one processor the load mode is not practical anymore, so there has to be ROM at 0000 to be executed after a reset. The EPROM is really 128k, but is addressed so that each CPU gets its own 16k portion.

      4000 - 4FFF Static RAM (4k):
      The main RAM is going to be banked, but for things like the stack there must also be a fixed RAM that can't be switched away. That's what this memory is for. Again it will be a 32k SRAM that provides each processor with its own 4k portion.

      5000 - 7FFF free (12k):
      Some processor may need an address space for memory mapped IO, graphics RAM or other purposes.

      8000 - FFFF paged RAM (32k):
      The 32k pages are shared between all processors. I have no MMU that could prevent two processors from selecting the same memory page. That's good if proper care is taken, but also problematic if not. The pages are selected with a simple 8 bit latch, so the computer can have 256 x 32k = 8 meg RAM.

      The first CPU will get all of the IO, simply because I don't see a way to access the registers of the controllers in a single clock cycle like the memory. At least we have no issues with multiple CPUs trying to do IO concurrently.

      The second CPU, I thought should take care of the graphics. The other CPUs simply issue graphics commands and provide graphics data in shared memory. For this we would need some kind of command queue and mechanisms to synchronize concurrent access, but those things would be for the OS to take care of.

      The other CPUs (two on each processor card) would be general purpose for parallel data crunching.

      As a first step, I would like to build a single board computer, practically the first CPU card plus some parts of the memory section to get those parts working before having to deal with the critical timing involved with multiplexing several processors on the same bus. This prototype alone would already be very much like the modernized Elf I proposed.

      --- In cosmacelf@yahoogroups.com, Andrew Wasson wrote:
      >
      > I agree, a modern ELF wouldn't be a bad project at all which is why Robert Armstrong built the ELF2K some time ago and my guess is that the ELF2K would be a good place to start your super computer.
      >
      > Out of the box, it runs the established low speed so that it is compatible with the Pixie chip but that isn't necessary if you don't use the Pixie graphics controller. I'd up the ante and put in a CDP1802BCE that runs 5Mhz.
      >
      > Graphics Card (composite): http://www.sparetimegizmos.com/Hardware/Elf2K_Accessories.htm#80%20Column%20Text%20Video%20Board
      >
      > * If you want a more advanced card, I'd look at what Quinn Dunkie is doing with her Veronica 6502 based computer. I've been following her video controller project with interest (about 5 posts on video lead up to this page): http://quinndunki.com/blondihacks/?p=1154
      >
      > PS2 Keyboard Interface: http://www.sparetimegizmos.com/Hardware/Elf2K_Accessories.htm#General%20Purpose%20I/O%20Card
      >
      > Disk Interface for mass storage: http://www.sparetimegizmos.com/Hardware/Elf2K_Accessories.htm#Disk,%20UART%20and%20RTC%20Board
      >
      > That should get you more than half way there. I think the hardware will be the easier side of the coin, the difficult part will be writing the OS and drivers to go along with it. I've got a bunch of 1804 chips looking for a home in a multi-processor ELF type computer.
      >
      > Andrew
      >
      >
      >
      >
      > On 2013-01-07, at 8:48 AM, Richard wrote:
      >
      > > How about designing a modernized Elf first? A common architechture to build and to write programs for.
      > >
      > > - Run the processor with full 3.2 MHz,
      > >
      > > - No more toggle switches. Just reset and start the computer from ROM.
      > >
      > > - 64 k memory or more (banked). 512k SRAM come on a single chip.
      > >
      > > - A modern keyboard interface. Parallel keyboards are hard to get by now, PS/2 also. USB would be good, but it's all SMD and 3.3 V.
      > >
      > > - Some kind of mass storage, or simple USB sticks (if we get USB working)
      > >
      > > - A sound chip (one from Yamaha should be relatively easy to get)
      > >
      > > - An adequate little OS in ROM, perhaps Elf OS can be adapted.
      > >
      > > - And last, a good graphics chip, perhaps a little better than the old MC6847. It should have a text mode and a nice color graphics mode (with an adequately low resolution). And, above all, we should design an interface to access graphics memory which does not slow the processor down too much.
      > >
      > > We could probably get all this onto a single board and get a quite capable 8 bit computer. And for those who want more: There still was that idea to build an Elf with up to 8 processors.
      > >
      > > --- In cosmacelf@yahoogroups.com, Jim Hall wrote:
      > > >
      > > > I just hope this kind of thinking is catching, I wrote this note to break us out of the same rut of thinking inside the box. So what if our 1802 has a on board cassette interface, or a small screen, the tought of something new gets the juices flowing, instead of rehashing idea's that have been rewritten too many times, over too many remakes and no fruit to harvest, other than hot air and nothing accomplished ? There is enough brain power in this group to see if it is possible, and then try it for real. I have seen the groups brain accomplish many things, but why not something NEW and reaching for a possible future ?
      > >
      > >
      >
    • Lee Hart
      ... There are lots of simple ways to network two or more micros, but IP/ethernet certainly isn t one of them. It s only simple in the sense that you can dump
      Message 140 of 140 , Jan 16, 2013
      • 0 Attachment
        On 1/15/2013 7:03 PM, Ray Po wrote:
        > Has anyone considered building a simple IP/ethernet interface to attach
        > 2 or more 1802's together and co-process?

        There are lots of simple ways to network two or more micros, but
        IP/ethernet certainly isn't one of them. It's only "simple" in the sense
        that you can dump your wallet on the table to buy some off-the-shelf
        board with software to do it all for you.

        All you really need is a simple serial port between them, and a little
        software. I've networked Parallax BASIC Stamps this way, and they are
        even dumber and slower than 1802's. :-)

        --
        The most dangerous enemy of a better solution is an existing one that
        is just good enough. -- Eric S. Raymond
        --
        Lee A. Hart, http://www.sunrise-ev.com/LeesEVs.htm
      Your message has been successfully submitted and would be delivered to recipients shortly.