IBM AT ISA bus card overclock

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

I have a machine that interfaces to a IBM AT 286 Xenix computer by
means of an ISA card. I would like to upgrade this machine to
something more recent like a 486 or pentium class processor inorder to
decrease compiling times when prorgams are written.

The problem I am encoutering is that the card is not recognized on the
new systems ISA bus. I'm running linux on the new system and lspci and
pnpdump show nothing. Is there anyways for me to overclock this cards
speed to match the new bus speed, or could I underclock the bus speed
of the new system so it will recognize the card?
9 answers Last reply
More about card overclock
  1. Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

    -----BEGIN PGP SIGNED MESSAGE-----
    Hash: SHA1

    In article <8406a76f.0411241619.447afe17@posting.google.com>,
    Josh <sinetific@gmail.com> wrote:
    >I have a machine that interfaces to a IBM AT 286 Xenix computer by
    >means of an ISA card. I would like to upgrade this machine to
    >something more recent like a 486 or pentium class processor inorder to
    >decrease compiling times when prorgams are written.
    >
    >The problem I am encoutering is that the card is not recognized on the
    >new systems ISA bus. I'm running linux on the new system and lspci and
    >pnpdump show nothing. Is there anyways for me to overclock this cards
    >speed to match the new bus speed, or could I underclock the bus speed
    >of the new system so it will recognize the card?

    A card that's old enough to have shipped in a 286 isn't going to support ISA
    Plug & Play, which was only introduced alongside Windows 95. Whatever
    resources (I/O, IRQ, etc.) the card is configured to use (whether through
    jumpers or a fixed configuration set by the hardware designer) it'll just go
    ahead and take for itself. lspci won't show anything different because the
    card you're trying to use is not a PCI card. I've never used pnpdump (ISA
    PnP was an evil hack by Microsoft that tended to work only when it felt like
    working), but it shouldn't list anything different either because while your
    card is an ISA card, it doesn't support ISA PnP.

    IIRC, ISA was never specced for speeds beyond 8 MHz. Some early 10-MHz
    machines might've overclocked the bus a bit, but I don't think there was
    ever a board produced for anything from the 386 on up that ran the ISA bus
    faster than 8 MHz. For a card pulled from a 286, ISA bus speed in a newer
    machine shouldn't be a problem. The only time I've had to underclock a
    machine to get a card working was when I needed to hook up the MFM
    controller and ST225 from a PC/XT to a newer machine to get DOS loaded onto
    the drive (my copy of DR DOS 6 is on 5.25" high-density floppies). I think
    the ISA bus speed remained constant; it was just the processor that was
    slowed from 233 MHz to somewhere around 90-100 MHz. I think I had to
    disable the L1 and L2 caches, too. That might've been an issue with the
    card's firmware not wanting to run right at faster speeds more than a
    bus-speed issue.

    _/_
    / v \ Scott Alfter (remove the obvious to send mail)
    (IIGS( http://alfter.us/ Top-posting!
    \_^_/ rm -rf /bin/laden >What's the most annoying thing on Usenet?

    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.2.4 (Linux)

    iD8DBQFBplJnVgTKos01OwkRAgPeAKCDCW1DdCxENEA1hggPk/xtR6DQRwCfcAu3
    +o4LHspZkSwF/mGtaRkaNDA=
    =DAai
    -----END PGP SIGNATURE-----
  2. Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

    sinetific@gmail.com (Josh) wrote :

    > ISA card
    > lspci and pnpdump show nothing

    lspci is not lsisa :)
    and the card might be configured by jumpers, not a p&pray

    what exactly is that card ? what OS was that 286 running ? have you got
    any drivers for it ? How are you going to drive it ? there is no way to
    simply "look" whats connected to ISA bus afaik beside BIOS
    initialisation.


    Pozdrawiam.
    --
    RusH //
    http://randki.o2.pl/profil.php?id_r=352019
    Like ninjas, true hackers are shrouded in secrecy and mystery.
    You may never know -- UNTIL IT'S TOO LATE.
  3. Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

    Josh <sinetific@gmail.com> wrote:
    > I have a machine that interfaces to a IBM AT 286 Xenix
    > computer by means of an ISA card. I would like to upgrade
    > this machine to something more recent like a 486 or pentium
    > class processor inorder to decrease compiling times when
    > prorgams are written.

    > The problem I am encoutering is that the card is not
    > recognized on the new systems ISA bus. I'm running linux on
    > the new system and lspci and pnpdump show nothing. Is there
    > anyways for me to overclock this cards speed to match the
    > new bus speed, or could I underclock the bus speed of the
    > new system so it will recognize the card?

    Well, I'll presume you have driver source and skills to migrate
    it to Linux. Otherwise, you'll have to migrate that Xenix.

    A couple of things you should check -- first, are the IRQs
    that card needs available on the ISA bus, or have they been
    stolen by PCI?

    Second, is that ISA bus running at satisfactory speed?
    Around the time of the 486/Pentium, PCI became primary
    and the ISA bus was hung off it through the Southbridge.
    This dramatically hurt performance, IIRC from ~5 MByte/s on
    a good ISA bus down to ~1 MByte/s typical through Southbridge.

    -- Robert
  4. Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

    The card is a not a well know card. It is an interface to a machine
    used for automatic testing of digital ICs. It has no jumpers or any
    settings at all onboard. It is running on SCO Xenix 2.2 System V. My
    skill on Xenix or quite lacking. I've been fumbling around with
    ancient versions of commands I know from linux trying to feel my way
    around the system. A manual or man pages would be a god send.

    I have a linux distro with linux-abi installed in the kernel waiting
    for me to transfer the digital testing software to it. I wanted to
    make sure the card would work before I undertook either:
    a) the uucp transfer of the software or
    b) the (5.25)floppy shuffle
    But I guess that isn't going to happen since it sounds like the card
    is software dependant. Mounting the drive in the new system is out of
    the question since the linux kernel module that claims to support
    sysv, xenix, and coherent file systems did not work for mounting the
    drive to the new system(unless I have a wrong syntax, but i double
    checked it with other people).
    Linux-abi claims support for xenix286 binaries and that is how I plan
    to run the old software on the new system. Unfortunately, I dont have
    the driver source. Do you think that this will be a major obstacle?
    The company that made the machine is no longer in buisness. I think
    the driver is built into the testing software.

    pnpdump seems to go through every address on the bus line to search
    for cards. nothing shows up from it, but possibly it is searching for
    pnp devices or some type of pnp response(I'm not that familiar with
    the inner workings of pnp).

    Is there anyway for me to find the IRQ of this card on the xenix
    system so I can make sure it is open on the new system? I will have to
    research this.

    There are also problems reading 5.25" floppies on the newer linux
    system.
  5. Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

    salfter@salfter.diespammersdie.dyndns.org (Scott Alfter) wrote :

    > IIRC, ISA was never specced for speeds beyond 8 MHz. Some early
    > 10-MHz machines might've overclocked the bus a bit, but I don't
    > think there was ever a board produced for anything from the 386 on
    > up that ran the ISA bus faster than 8 MHz

    well I'v got ADAPTEC SCSI ISA card that can work with ISA up to 12MHz


    Pozdrawiam.
    --
    RusH //
    http://randki.o2.pl/profil.php?id_r=352019
    Like ninjas, true hackers are shrouded in secrecy and mystery.
    You may never know -- UNTIL IT'S TOO LATE.
  6. Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

    On Thu, 25 Nov 2004 21:45:12 GMT, salfter@salfter.diespammersdie.dyndns.org
    (Scott Alfter) wrote:

    >-----BEGIN PGP SIGNED MESSAGE-----
    >Hash: SHA1
    >
    >In article <8406a76f.0411241619.447afe17@posting.google.com>,
    >Josh <sinetific@gmail.com> wrote:
    >>I have a machine that interfaces to a IBM AT 286 Xenix computer by
    >>means of an ISA card. I would like to upgrade this machine to
    >>something more recent like a 486 or pentium class processor inorder to
    >>decrease compiling times when prorgams are written.
    >>
    >>The problem I am encoutering is that the card is not recognized on the
    >>new systems ISA bus. I'm running linux on the new system and lspci and
    >>pnpdump show nothing. Is there anyways for me to overclock this cards
    >>speed to match the new bus speed, or could I underclock the bus speed
    >>of the new system so it will recognize the card?
    >
    >A card that's old enough to have shipped in a 286 isn't going to support ISA
    >Plug & Play, which was only introduced alongside Windows 95. Whatever
    >resources (I/O, IRQ, etc.) the card is configured to use (whether through
    >jumpers or a fixed configuration set by the hardware designer) it'll just go
    >ahead and take for itself. lspci won't show anything different because the
    >card you're trying to use is not a PCI card. I've never used pnpdump (ISA
    >PnP was an evil hack by Microsoft that tended to work only when it felt like
    >working), but it shouldn't list anything different either because while your
    >card is an ISA card, it doesn't support ISA PnP.

    IIRC the M$ specs for ISA PnP simply stated that the PnP ISA card had to
    have one alternative for each of IRQ, I/O address and DMA, IOW two of each
    resource it required and that it be capable of switching between them. Yes
    it was a hack and it caused as much trouble as benefit... and many BIOSs
    did not handle things right anyway.

    Rgds, George Macdonald

    "Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
  7. Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

    On 24 Nov 2004 16:19:57 -0800, sinetific@gmail.com (Josh) wrote:

    >I have a machine that interfaces to a IBM AT 286 Xenix computer by
    >means of an ISA card. I would like to upgrade this machine to
    >something more recent like a 486 or pentium class processor inorder to
    >decrease compiling times when prorgams are written.

    Just to be clear: it's the IBM AT 286 that you want to replace - yes?

    >The problem I am encoutering is that the card is not recognized on the
    >new systems ISA bus. I'm running linux on the new system and lspci and
    >pnpdump show nothing. Is there anyways for me to overclock this cards
    >speed to match the new bus speed, or could I underclock the bus speed
    >of the new system so it will recognize the card?

    Do you know which resources the ISA card uses - IOW does it have mapped
    memory... on top of the usual IRQ & I/O address?... and does it use DMA?
    Also what is this "new system"? Can you not tell what this ISA card is or
    what it does? I wish you'd been more specific. Depending on the resources
    required and the target computer, there are several things which could go
    wrong.

    There were systems produced, in the 386/486 timeframe which used chipsets
    which did not work with many cards which had mapped memory on them - the
    memory bus was slightly different from IBM's original AT standard and
    addresses were latched with edge triggered D flip-flops... if I'm
    remembering things right.

    Overclocking the ISA card is out but there were many mbrds produced in the
    early PCI days which had speed programmable ISA buses: typically, in the
    BIOS Setup you could set the speed ratio of PCI:ISA to 4:1(standard 8MHz),
    3:1, 5:2 etc. The other thing to watch is that with PnP BIOSs, there was
    often the option to "reserve" IRQs and I/O addresses in BIOS Setup, which
    told the BIOS to block their use for PnP and allow them to be dedicated to
    a specific card on the ISA bus.

    Rgds, George Macdonald

    "Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
  8. Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

    On 24 Nov 2004 16:19:57 -0800, sinetific@gmail.com (Josh) put finger
    to keyboard and composed:

    >I have a machine that interfaces to a IBM AT 286 Xenix computer by
    >means of an ISA card. I would like to upgrade this machine to
    >something more recent like a 486 or pentium class processor inorder to
    >decrease compiling times when prorgams are written.
    >
    >The problem I am encoutering is that the card is not recognized on the
    >new systems ISA bus. I'm running linux on the new system and lspci and
    >pnpdump show nothing. Is there anyways for me to overclock this cards
    >speed to match the new bus speed, or could I underclock the bus speed
    >of the new system so it will recognize the card?

    A photo of the card may be of help. Some older jumperless non-PnP ISA
    cards used a configuration utility to set up the card's resources. My
    old 486 machine has a sound card of this type. I also have a non-PnP
    jumperless NE2000 compatible NIC. In the case of the NIC, the config
    utility was used to intially set up the card's resources, after which
    they would be written to a serial EEPROM chip. In the case of the
    sound card, the resources were not stored on the card but were
    initialised at bootup by means of a statement in the autoexec.bat
    file.

    BTW, I also have a specialised ISA interface card for a device
    programmer and IC tester (Sunshine Expro 60). I keep my old 486 box
    alive because this card does not work reliably in faster systems, even
    after experimenting with its wait state jumpers.


    - Franc Zabkar
    --
    Please remove one 's' from my address when replying by email.
  9. Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

    On 26 Nov 2004 02:52:08 -0800, sinetific@gmail.com (Josh) put finger
    to keyboard and composed:

    >The card is a not a well know card. It is an interface to a machine
    >used for automatic testing of digital ICs. It has no jumpers or any
    >settings at all onboard.

    It may have some hardwired etches. For example, there may be an etch
    that configures the card as primary or secondary to allow for two
    cards in the one box.

    >It is running on SCO Xenix 2.2 System V. My
    >skill on Xenix or quite lacking. I've been fumbling around with
    >ancient versions of commands I know from linux trying to feel my way
    >around the system. A manual or man pages would be a god send.
    >
    >I have a linux distro with linux-abi installed in the kernel waiting
    >for me to transfer the digital testing software to it. I wanted to
    >make sure the card would work before I undertook either:
    >a) the uucp transfer of the software or
    >b) the (5.25)floppy shuffle
    >But I guess that isn't going to happen since it sounds like the card
    >is software dependant. Mounting the drive in the new system is out of
    >the question since the linux kernel module that claims to support
    >sysv, xenix, and coherent file systems did not work for mounting the
    >drive to the new system(unless I have a wrong syntax, but i double
    >checked it with other people).

    Is the drive an IDE type? Older 286 machines used a special MFM/RLL
    interface card, for example. Installing this card into a machine with
    its own IDE ports would not work. Sorry if this was stating the
    obvious.

    >Linux-abi claims support for xenix286 binaries and that is how I plan
    >to run the old software on the new system. Unfortunately, I dont have
    >the driver source. Do you think that this will be a major obstacle?
    >The company that made the machine is no longer in buisness. I think
    >the driver is built into the testing software.

    That's the case with my Expro 60 device programmer/tester interface
    card. The card's resources are invisible to Windows.

    >pnpdump seems to go through every address on the bus line to search
    >for cards. nothing shows up from it, but possibly it is searching for
    >pnp devices or some type of pnp response(I'm not that familiar with
    >the inner workings of pnp).
    >
    >Is there anyway for me to find the IRQ of this card on the xenix
    >system so I can make sure it is open on the new system? I will have to
    >research this.

    If at all possible, please post a photograph or scan of both sides of
    your card on your web space. The part numbers of the various chips may
    give a clue. Some people who watch comp.sys.ibm.pc.HARDWARE.CHIPS are
    actually hardware people. ;-)

    For example, knowing whether the card occupies one or two connectors
    would define it as an 8 bit or 16 bit card. An 8 bit card would be
    limited to IRQs 3,4,5,7 and 9. If the card contains no intelligent
    chips or chipsets (eg microcontrollers + RAM + ROM), then that would
    suggest that all the resources would come under software control. If
    the card has a PAL IC, then that is where I would be looking to
    determine its IO address. My Expro 60's interface card is polled
    rather than interrupt driven. You can determine whether this is the
    case for your card by inspecting the IRQ pins on the ISA fingers.
    Unused IRQs would have no copper traces leading to their respective
    ISA pins. IRQs 3,4,5,7,9 connect to pins B25, B24, B23, B21, and B4.
    IRQs 10,11,12,15 connect to D3, D4, D5, and D6. You can do the same
    for the DMA pins.

    >There are also problems reading 5.25" floppies on the newer linux
    >system.

    What kind of problem? Can you format a fresh diskette on this drive?
    Have you tried using DOS to test your floppy drive?


    - Franc Zabkar
    --
    Please remove one 's' from my address when replying by email.
Ask a new question

Read More

CPUs Overclocking Bus IBM