1. This forum section is a read-only archive which contains old newsgroup posts. If you wish to post a query, please do so in one of our main forum sections (here). This way you will get a faster, better response from the members on Motherboard Point.

Difference between IDPROMs

Discussion in 'Sun Hardware' started by Wes Groleau, Mar 24, 2012.

  1. Wes Groleau

    Wes Groleau Guest

    References I found said listed IDPROM part numbers for Sun Ultra 10 as
    525-1430, 525-1817, and f100-6889. Are they really three different
    things? Checking prices, one place has a web page for each, with the
    only difference being the part number, and a price of $38 for each.

    Another place says original price $95 but they'll give me incredible
    savings and sell it to me for fifty-some (same for all three numbers)
    A third place give a different price for each, saying original price is
    $250-three something, but they'll give me an incredible bargain by
    letting me have it for $120-160.

    Most of the places won't tell you a price without enough info for a
    sales person to bug you.

    Anyway, I decided to take a chance on a used one for $20 but still
    curious whether there is any difference between the three numbers.

    Oh, by the way, if the Ultra ten box is assembled and plugged in, but
    "OFF", is there still a trickle path to keep the IDPROM charged? Is
    there any chance that if plugged in long enough, I could revive and
    reprogram it instead of replacing it?

    Wes Groleau

    Always listen to experts. They'll tell you
    what can't be done and why. Then do it.
    — Robert A. Heinlein
    Wes Groleau, Mar 24, 2012
    1. Advertisements

  2. Wes Groleau

    DoN. Nichols Guest

    Two differences possible.

    1) The information already written in there may be different
    such as the upper bits of the HOSTID, which defines the
    particular version of the system. Since the system was
    made with several different CPU speeds, it may be this, a
    different HOSTID upper part for different CPU speeds.

    2) The components which make up the IDPROM -- a merge of a
    CMOS RAM, a time-of-day clock chip, and a cell (e.g. component
    of a battery), which is used to both preserve the contents of
    the RAM when the system is powered down, and to power the clock
    chip. (The clock chip is halted when a new one is received, and
    when the system powers up, it starts the chip keeping time.)
    This makes them last longer on the shelf -- and if there were a
    utility included to set the halt bit again, it would allow a
    system to go into storage for longer times without problems.

    But -- Sun used a bug in the original chip to accomplish
    something -- I don't know what. The new chips no longer have
    that bug, and the system requires a different ROM in it to deal
    with the fact that it no longer has the bug to work with. So,
    newer chips won't work on older systems. At a guess, that may
    be the difference between the 525-???? ones and the f100-6889.

    O.K. A web search found this site:


    which shows that all the versions listed for the Ultra-5 and Ultra-10
    are the same physical chip --- 48T59 NVRAM, so it must be differences in
    the information included -- likely CPU speed and perhaps other changes
    in the system board.
    Used ones will have been run for some unknown time, and it is
    unlikely that anyone who pulled them from the systems bothered entering
    the fcode to re-write the halt bit, so they will continue aging on the
    shelf. They are better in a system which is powered up and running.
    I don't know whether there is power to keep the clock running.
    At best, if there is, it will keep the clock from discharging the cell
    built in -- but *cannot* recharge it, as it is a primary cell (not
    No! However, there are web sites which will tell you how to
    perform surgery on the chip housing to mount an external cell in place
    of the now dead one. If you don't have the hostid and the ethernet MAC
    address from before noted down, you will have to jump through hoops to
    generate bogus (but acceptable) new ones which won't confuse the system.

    Note that one of the SPARC based semi-clones, the Solbourne
    S4000 and S4000DX did things *right*. The HOSTID and ethernet MAC
    address were burned into a small bipolar ROM, and the NVRAM/clock chip
    was kept powered by a coin cell in a holder on the sytem board. You
    could lose NVRAM settings, but nothing so critical as the HOSTID and MAC

    Now -- some later machines, such as the Sun Blade 1000/2000 and
    Sun Fire 280R use a SEEPROM (Serial EEPROM) which does not need power to
    retain its data, and a separate clock chip. The web site says that the
    Sun Blade 1500 and 2500 use the same SEEPROM, with four different part
    numbers for the four speeds of systems -- and they don't bother
    mentioning the Sun Fire 280R, which uses the same system board as the
    Sun Blade 1000/2000.

    And machines like the Sun Fire V120 have the HOSTID and ethernet
    MAC address on a front-panel plug-in card, so you can transfer license
    code to a replacement system while you swap the disk drives to keep
    going as quickly as possible.

    So -- Sun finally got that right. But too late, I fear, since
    Oracle has taken over the helm.

    Good Luck,
    DoN. Nichols, Mar 25, 2012
    1. Advertisements

  3. Wes Groleau

    Wes Groleau Guest

    The "hoops" of typing FORTH commands seem less daunting to me than
    prying open an NVRAM and soldering in a watch battery. :)

    MAC is no big deal, since it will not be directly connected to the
    Internet and there are no other Suns in the house.

    Is hostid an issue when I intend to reload with Solaris 10 ?

    Wes Groleau

    Always listen to experts. They'll tell you
    what can't be done and why. Then do it.
    — Robert A. Heinlein
    Wes Groleau, Mar 25, 2012
  4. No, hostid is only used for a few 3rd party software licensing
    (ie. flexlm based systems and the like).
    Solaris doesn't use it in any fashion.

    The MAC address stays on your gateway router anyway. It isn't
    broadcast beyond that. So even if you duplicate a MAC address two hops
    away, it won't matter.
    Doug McIntyre, Mar 25, 2012
  5. Maybe so, but ... I wouldn't bet on those said systems staying two hops
    apart forever ... aren't there any private MACs similar to RFC1918 IPs?
    Volker Borchert, Mar 25, 2012
  6. Wes Groleau

    Wes Groleau Guest

    Thanks. good to know.
    That's what I thought, with about five percent uncertainty. :)
    Wes Groleau, Mar 25, 2012
  7. Wes Groleau

    Wes Groleau Guest

    The probability that I will sell this Sun to someone who has other Suns
    is low. The probability that one of his other Suns will have the MAC I
    choose is even lower.

    I don't know the probability that he'll ignore my warning to check for
    conflict, but by that time, I won't care.

    Wes Groleau

    Even if you do learn to speak correct English,
    whom are you going to speak it to?
    — Clarence Darrow
    Wes Groleau, Mar 26, 2012
  8. Wes Groleau

    Wes Groleau Guest

    That wouldn't help. RFC1918 provides groups of addresses to use within
    a LAN. If another LAN adds a machine having the same RFC1918 IP as one
    they already had, the effects are similar to adding a machine with a MAC
    that they already have.

    Wes Groleau

    Always listen to experts. They'll tell you
    what can't be done and why. Then do it.
    — Robert A. Heinlein
    Wes Groleau, Mar 26, 2012
  9. Wes Groleau

    DoN. Nichols Guest

    You still have to do the hoops even with a new chip -- unless it
    is specifically programmed to be a Ultra-10 chip (right HostID). With
    the old chip, as soon as you turn it off, you will lose whatever you
    have put into the NVRAM once the chip reaches "that age".
    O.K. Until someone else gets the system.
    Yes. The first nybble of the HostID -- or maybe the first byte
    -- defines what kind of system it is. The OS uses that to determine the
    interfaces which should be built in, and the addresses at which they are
    located. With the wrong HostID, you can wind up with the OS not being
    able to find critical things. (For example, most UltraSPARC machines
    expect to talk to disk drives through either SCSI or Fibre Channel. The
    Ultra-5 and Ultra-10 are uncommon in that they used IDE (PATA) drives
    instead. The lower part of the HostID doesn't matter -- unless you have
    licensed software keyed to the hostid. :)

    This URL has ways of programming the values -- but it is a bit
    old, since it does not seem to mention any of the UltraSPARC systems.


    And this one has a program which is shown being used on an
    Ultra-5/Ultra-10 (They use the same system board -- only the case is


    Make the first two nybbles of the hostid match what is shown,
    and pick your own remaining digits -- ideally somewhat high in the range
    so you don't duplicate another system.

    Good Luck,
    DoN. Nichols, Mar 26, 2012
  10. Wes Groleau

    DoN. Nichols Guest

    The usual solution is to pick up an old ethernet PC-board for
    some old system (probably a PC with an ISA bus so it is less likely to
    be needed again), find the MAC address used by that board, and use it,
    and pull and destroy the small ROM which defines the MAC address to the
    board. This insures that there is no duplication.

    DoN. Nichols, Mar 26, 2012
  11. Wes Groleau

    Hank Guest

    On all of the Sparc Ultras, (sun4u), the NVRAM is a Signetics
    M48T59-70PC1. Available from Memoryx unprogrammed.

    The first byte on an Ultrasparc is 80.
    The hostid and MAC are tied together. On the later Suns, the original
    ID is on a printed label on the original NVRAM. If programming a bare
    NVRAM, just use that number. When I do this, I lift off the label and
    stick it on the new NVRAM with Pliobond. I've probably done 10 or 15
    of these over the years, and it's a pretty straightforward job.

    Can't recall the exact details now, but I think it was "filch an NVRAM
    from a scrap U1 for a bare U5/10 motherboard." As I recall, I built
    the system up and on bootup, it hollered that the checksum was
    incorrect until I rebooted it with the keyboard default reset. I
    thought I'd have to reprogram it, but as I recall, I didn't.
    If the user is replacing the original NVRAM on an Ultra, the data is
    on the label, as I mentioned above. Not true for earlier systems like
    the SS5/20. As long as the first byte is 80, the other digits
    generally don't matter, as long as they don't conflict with the MACs
    in the local system.

    The arp -a command will show the MAC addresses of everything a system
    has communicated with. On this Ultra 60, it begins 08:00:20. Note
    that the 80 mentioned earlier appears as 08 in the MAC.

    Hank, Mar 26, 2012

  12. I don't believe that has been relavent inside Solaris or the OBP for
    quite some time. Like before it was even called Solaris..

    Maybe if you had some Sun4 machines..

    Hmm, maybe I should disassemble the OBP in the sun4 I have in the
    other room..
    Doug McIntyre, Mar 26, 2012
  13. Wes Groleau

    Wes Groleau Guest

    Right, but I'd rather do that than hardware hacking.
    Not likely to happen. Even less likely someone who gets it will already
    have the MAC I picked out of thousands of possible values.
    But if it will make everyone happy, I could forget the Sun prefix
    and copy a MAC from one of the not-worth-fixing things I haven't
    thrown away yet. Or in a slim chance of revenge, use the Mac of
    of my stolen iPhone. :)
    Ah, someone else said it doesn't matter, but now that you mention it,
    I remember seeing a chart of what hostid to use.
    I think that's the one I used before. Old, but what it was missing, I
    found in manuals.
    Wes Groleau, Mar 26, 2012
  14. Not surprising, as 8:0:20 is one of the vendor prefixes assigned to Sun.
    Volker Borchert, Mar 27, 2012
  15. Wes Groleau

    Hank Guest

    Hank, Mar 28, 2012
  16. Wes Groleau

    DoN. Nichols Guest

    [ ... ]

    [ ... ]
    In both the Sun Fire V120, and the Sun Blade 2000, the first
    byte is 83, not 80. (Those are both Uttra-II CPUs, IIRC.) I don't have
    any Ultra-5 or Ultra-10 systems booted at the moment to check.

    DoN. Nichols, Mar 28, 2012
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.