Shop OBEX P1 Docs P2 Docs Learn Events
TTL computer — Parallax Forums

TTL computer

icepuckicepuck Posts: 466
in General Discussion
https://gigatron.io/

Just found this while trolling youtube. A computer using ttl logic in place of a cpu.
-dan

Comments

  • Heater.Heater. Posts: 21,230
    Yes, a brilliant project.

    If we believe what it says then the TTL computer is faster and smaller than the 8 bit machines that were all the rage. Like the Apple 8 bitters.

    If so, why did anyone use an 8 bit CPU back in the late 1970's ?

  • Cluso99Cluso99 Posts: 18,069
    Heater. wrote: »
    Yes, a brilliant project.

    If we believe what it says then the TTL computer is faster and smaller than the 8 bit machines that were all the rage. Like the Apple 8 bitters.

    If so, why did anyone use an 8 bit CPU back in the late 1970's ?
    1. Less components = smaller or less pcbs
    2. Cheaper
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    There seems like a dearth of components on that board for what it's stated to do. I wonder if they're using AM29xx bit-slice ICs.

    -Phil
  • Heater.Heater. Posts: 21,230
    There is the thing Cluso, Looking at that TTL computer it seems to be smaller than the equivalent early 8 bitters in terms of board space. And better performing too.

    Given the price of 8 bit CPU's at the time I can imagine that board of TTL would have be cheaper as well.

    I don't know what to make of the guys claims.

    Of course, what the microprocessor sold us at the time was ease of programming. I suspect that TTL machine does not win there.

  • jmgjmg Posts: 15,173
    Phil Pilgrim (PhiPi) wrote: »
    There seems like a dearth of components on that board for what it's stated to do. I wonder if they're using AM29xx bit-slice ICs.
    The clock speed is modest, at 6.25MHz, and it looks to swallow a good chunk of Logic/opcode decode into the large/wide 64kx16 ROM.
    It also has just 7 core opcode groups, with mode modifiers, so code efficiency is not great.
    If it can run old games, that's not doing too bad...

  • Heater.Heater. Posts: 21,230
    I don't see any AM29xx bit-slice ICs in there. They are huge. The 2901 was a 40 pin package and the 2903 a 48 pinner. That is per 4 bit slice. Nobody used those at the time unless they were building a 16 bit, or more, machine.

    I do wonder if those big ROMs were available at the time for the decode logic.

    7 core opcode groups might make it a RISC machine. Ahead of its supposed time!

  • kwinnkwinn Posts: 8,697
    Heater. wrote: »
    I don't see any AM29xx bit-slice ICs in there. They are huge. The 2901 was a 40 pin package and the 2903 a 48 pinner. That is per 4 bit slice. Nobody used those at the time unless they were building a 16 bit, or more, machine.

    I do wonder if those big ROMs were available at the time for the decode logic.

    7 core opcode groups might make it a RISC machine. Ahead of its supposed time!

    IIRC there were no standard programmable roms of that size or width at the time. There were small (~1K - 256x4, 128x8) one time proms around. Of course things were changing so fast that what was available changed from month to month so "at the time" had to be a pretty specific year and month.
  • Dave HeinDave Hein Posts: 6,347
    Looking at posts on Hackaday, the ALU consists of 2 74HCT283, which are 4-bit adders, and 8 74HCT153, which are dual 4-to-1 muxes. The ALU implements and, or, xor, add, sub and pass-thru. They considered using 2 74181 ALU chips, but they wanted to use simpler chips for the project.
  • Cluso99Cluso99 Posts: 18,069
    Heater. wrote: »
    There is the thing Cluso, Looking at that TTL computer it seems to be smaller than the equivalent early 8 bitters in terms of board space. And better performing too.

    Given the price of 8 bit CPU's at the time I can imagine that board of TTL would have be cheaper as well.

    I don't know what to make of the guys claims.

    Of course, what the microprocessor sold us at the time was ease of programming. I suspect that TTL machine does not win there.
    I was building CPU boards back then. I can assure you that the part costs (including pcbs which were v.expensive) and assembly/test costs were much more expensive than the 6802 / 6502 of the day. Even when I used a pair of $150 MC68705P3S they were way less cost thean the board(s) of logic they replaced!!

  • DigitalBobDigitalBob Posts: 1,513
    I would like to see more detail like schematics etc.
  • TorTor Posts: 2,010
    It's a pipelined RISC system. That does seem to count for something. I actually have all the parts from the parts list in the drawer, or nearly so. So I wish they would sell bare PCBs and not only full kits.
  • The_MasterThe_Master Posts: 200
    Let me take the occasion to gauge interest in a potential kit product:

    Would there be any interest in a microprocessor kit that consisted entirely of TO-92 transistors? Apx 300 transistors to be soldered. They are arranged in a 17 x 17 monolithic grid. (A handful of resistors hidden on the underside of the board)

    Of course, like the monster6502, there are through-hole LED's intersped among the transistors.

    Obviously a 300 transistor microprocessor would be very, very simple: 4 bit, 5 instructions. But enough to run usefull demo programs. A socket on the board to plug in an EPROM with your program.
  • jmgjmg Posts: 15,173
    The_Master wrote: »
    Let me take the occasion to gauge interest in a potential kit product:

    Would there be any interest in a microprocessor kit that consisted entirely of TO-92 transistors? Apx 300 transistors to be soldered. They are arranged in a 17 x 17 monolithic grid. (A handful of resistors hidden on the underside of the board)

    Of course, like the monster6502, there are through-hole LED's intersped among the transistors.

    Obviously a 300 transistor microprocessor would be very, very simple: 4 bit, 5 instructions. But enough to run usefull demo programs. A socket on the board to plug in an EPROM with your program.

    You would need to be masochist to solder than many TO92, to end up with something that can't even tell the time...
  • tomcrawfordtomcrawford Posts: 1,126
    The ROM and the RAM do not occupy the same address space. The native CPU can execute instructions from the ROM only. This means that if you want to write code for the native machine you need to be able to program a ROM or else have a ROM emulator.
  • frank freedmanfrank freedman Posts: 1,983
    You would need to be masochist to solder than many TO92, to end up with something that can't even tell the time...

    Actually, look up CP-642B for a look what was done with only transistors.
  • Heater.Heater. Posts: 21,230
    edited 2018-04-04 16:13
    That CP-642B looks great.

    Back in 1980 when I started work for Marconi Radar they still had a Marconi Myriad minicomputer working in a back room. Built from transistors.

    product-96548.jpg

    Looks like it was decommissioned the year after I left:
    http://www.computinghistory.org.uk/det/43525/Marconi-Myriad-Photographs/

Sign In or Register to comment.