640K should be enough for anyone... - Minix

This is a discussion on 640K should be enough for anyone... - Minix ; Back in the day we used to have this weird 1M memory limit, a silly hack from a silly processor that hacked 4 bits on top of a 16 bit address space by using a 16 bit offset in a ...

+ Reply to Thread
Results 1 to 5 of 5

Thread: 640K should be enough for anyone...

  1. 640K should be enough for anyone...

    Back in the day we used to have this weird 1M memory limit, a silly
    hack from a silly processor that hacked 4 bits on top of a 16 bit
    address space by using a 16 bit offset in a 4 bit segment controlled
    by a register. As amazing as it seems, this hack not only worked, but
    also allowed an OS to run in 16-bit protected mode running a full
    graphical windowing system and multiple programs at once... in 640K of
    RAM, with video and memory mapped IO above that limit.

    I've thought about making Linux run in that space, but no... need like
    2 megs of RAM, getting an initrd uptakes more. What about BSD?
    Probably not. What else could I do?

    Minix 3 runs in what? 20k for the core kernel (4 pages?!), maybe 70k
    overall with all the systems running? Maybe a small windowing
    system... something like kdrive might get X running in under a meg.
    Just need a tiny window system and some desktop utilities.

    This is just hypothetical for me right now, but what's the possibility
    of getting Minix3 at a shell in say 100k? GUI in 1M or so? Running
    in VMware or VirtualBox usably in under 4M?

    My thought process is as such:

    * Boot Minix 3 in VMware or VirtualBox in 1M of memory, enough for a
    bootloader to get the core sytem excutable and a hard disk driver up.
    * Get Minix 3 to a small shell. BASH is too big, busybox + uClibc
    make 1M + 250k file size and won't use that much memory at once so...
    * Get kDrive or a more minimal X running, inside 2M
    * IceWM-like window manager, file browser, text editor, small web
    browser like Dillo, inside 4M

    It doesn't seem doable in this day and age to get a real OS working
    inside 640k or (so we're 32-bit now) 1280k; but I am still curious as
    to the expected possibility of making a true, unix-like OS with a real
    X11R6 system run inside such tight constraints. Windows 3.1 runs in
    2M and 3.11 wants 3M, so why not.

    Thoughts?

  2. Re: 640K should be enough for anyone...

    --{ John Moser a plopé ceci: }--

    > Thoughts?


    AmigaOS run in 256k.

    --
    Ca sert pas ā grand chose de dire "wiki" en sautant comme un cabri ;->

  3. Re: 640K should be enough for anyone...

    > It doesn't seem doable in this day and age to get a real OS working
    > inside 640k or (so we're 32-bit now) 1280k; but I am still curious as
    > to the expected possibility of making a true, unix-like OS with a real
    > X11R6 system run inside such tight constraints. Windows 3.1 runs in
    > 2M and 3.11 wants 3M, so why not.


    Just X11R6 is larger, especially in Minix due to lack of VM. I think it
    would never fit in 640K, you would need a (much) smaller window manager.

    --
    With kind regards,
    Erik van der Kouwe

  4. Re: 640K should be enough for anyone...

    Self modifying code - that, plus a virtual swap area - Together, it might
    just work... Would not want to debug it, though...

    ;-)


    "John Moser" wrote in message
    news:17c3a602-478e-4f1a-9491-d7b6276764ba@z72g2000hsb.googlegroups.com...
    > Back in the day we used to have this weird 1M memory limit, a silly
    > hack from a silly processor that hacked 4 bits on top of a 16 bit
    > address space by using a 16 bit offset in a 4 bit segment controlled
    > by a register. As amazing as it seems, this hack not only worked, but
    > also allowed an OS to run in 16-bit protected mode running a full
    > graphical windowing system and multiple programs at once... in 640K of
    > RAM, with video and memory mapped IO above that limit.
    >
    > I've thought about making Linux run in that space, but no... need like
    > 2 megs of RAM, getting an initrd uptakes more. What about BSD?
    > Probably not. What else could I do?
    >
    > Minix 3 runs in what? 20k for the core kernel (4 pages?!), maybe 70k
    > overall with all the systems running? Maybe a small windowing
    > system... something like kdrive might get X running in under a meg.
    > Just need a tiny window system and some desktop utilities.
    >
    > This is just hypothetical for me right now, but what's the possibility
    > of getting Minix3 at a shell in say 100k? GUI in 1M or so? Running
    > in VMware or VirtualBox usably in under 4M?
    >
    > My thought process is as such:
    >
    > * Boot Minix 3 in VMware or VirtualBox in 1M of memory, enough for a
    > bootloader to get the core sytem excutable and a hard disk driver up.
    > * Get Minix 3 to a small shell. BASH is too big, busybox + uClibc
    > make 1M + 250k file size and won't use that much memory at once so...
    > * Get kDrive or a more minimal X running, inside 2M
    > * IceWM-like window manager, file browser, text editor, small web
    > browser like Dillo, inside 4M
    >
    > It doesn't seem doable in this day and age to get a real OS working
    > inside 640k or (so we're 32-bit now) 1280k; but I am still curious as
    > to the expected possibility of making a true, unix-like OS with a real
    > X11R6 system run inside such tight constraints. Windows 3.1 runs in
    > 2M and 3.11 wants 3M, so why not.
    >
    > Thoughts?




  5. Re: 640K should be enough for anyone...

    On Jun 5, 3:12*am, "Erik van der Kouwe" few.vu.nl>
    wrote:
    > Just X11R6 is larger, especially in Minix due to lack of VM. I think it
    > would never fit in 640K, you would need a (much) smaller window manager.


    True, but as I said, Windows 3.1 wanted 2M and 3.11 wants 3M. 640K is
    an amusing enough goal due to the famous (mis?)quote; 2M is
    demonstratably possible in some plane of existence. X11 is bigger, so
    who is going to rewrite X11? That's a big undertaking, and will
    probably in itself ensure this stays a hypothetical proposition.

    On Jun 5, 12:25 pm, "R.A. Nagy" wrote:
    > Self modifying code - that, plus a virtual swap area - Together, it might
    > just work... Would not want to debug it, though...
    >


    Eww :P


    On Jun 4, 9:56 pm, "Thierry B." wrote:
    > --{ John Moser a plopé ceci: }--
    >
    > > Thoughts?

    >
    > AmigaOS run in 256k.


    Check mate, you win.

+ Reply to Thread