• Another myth-legend debunked [Corvette on 8580 details]

    From Louis Ohland@ohland@charter.net to comp.sys.ibm.ps2.hardware on Wed Oct 2 08:32:07 2024
    From Newsgroup: comp.sys.ibm.ps2.hardware

    Long, convoluted story SNIPped.

    Katherine Rohl was probing her 8580-Axe, and I brought up the "fact"
    that the Corvette would not configure on an 8580.

    While I was listing the RTC/SRAM/NVRAM configurations for all PS/2s, I
    noticed that the -Axx system board uses >8KB NVRAM<... but according to myth-legend, the Corvette needs 8K NVRAM... Huh, Way To the Future?

    Unlike me, Katherine fired up some painful probulation interrogation
    devices, attached them to very sensitive portions of her 8580, and
    extracted unknown information...

    The Corvette lacks a SCSI BIOS. It needs either another SCSI controller
    with a SCSI BIOS, -OR- an IML BIOS with the SCSI BIOS baked in.

    Katherine Rohl on the Corvette in an 8580: --------------//

    Also, the problem with the Corvette adapter in the Model 80 was obvious
    now that I've realized it: There's no SCSI BIOS. It presumably goes in
    the empty PLCC socket. The mapping in SC is just for its *address
    space.* I guess it's expecting to be run in an IML system which provides
    a SCSI BIOS to place in that address space.

    I don't have any IML systems, though - I'm going to experiment with
    modifying the Corvette ADP to *not* unload Spock's SCSI BIOS and see if
    that does anything.

    My thinking is that you could probably come up with a SCSI BIOS pretty
    easily for the Corvette. Since it’s in the same family as Spock (the ADP
    is nearly identical as well, I need to document the differences) I bet
    it was another one of those IBM “if you want SCSI-2, buy a 90 or better” bonehead decisions to protect server sales.

    The ADP (all of them, Tribble/Spock too) defers to IML if the system’s
    BIOS feature bits declare that it is an IML system with SCSI BIOS
    present. Since that’s not the case in a Model 80, it falls back to
    expecting the BIOS to be on the adapter and declaring its upper memory
    range. But there’s no device present so it just returns an open bus. Declaring a memory range and running the option ROM are two unconnected
    steps in the boot sequence - POST searches for option ROMs like a PC,
    not at a table of declared POS memory ranges.

    The easiest thing would be if you have a Corvette card in something that
    does work with it, identify and dump the IML SCSI BIOS range in upper
    memory and compare it to the known 1991 SCSI BIOS.
    --- Synchronet 3.20a-Linux NewsLink 1.114