• Hot gender-bender action! [MCMaster on RS/6000]

    From Louis Ohland@21:1/5 to All on Mon Feb 7 08:44:00 2022
    The biggest issue will be to recognize the MCMaster. AIX does not use
    ADFs, but an ODM. The RS/6000 must be successfully IML'd, and the
    "ADFUTIL" run in order to process the MCMaster's ADF, and IIRC there is
    an Innnn.ADF as well.

    Without successful addition of the MCMaster to the ODM, the MCMaster
    will NOT be recognized and activated during set-up. All it will do is
    continue to dream in the endless sleep....

    Then, there is the torrid gender-bender confusion of an Intel CPU on a
    POWER system. It could be done, [68K, Transputer] other CPUs have been
    run in PS/2s, but they had smart folks writing code to enable it.

    POWER is big Endian, 486 is little Endian. IMHO, PS/2s are single image systems, biased toward intel CPUs. RS/6000s are dual image systems,
    where the MCA is accessed through a "bridge". POWER systems use markedly different IML, memory controllers, and it's all run by AIX.

    So... in short, it is Possumble, but you need some pretty heavy
    AIX-chops to get it to work.

    An aside, even the lowly -320 could probably curb-stomp a 486.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From schimmi@21:1/5 to Louis Ohland on Mon Feb 7 14:50:15 2022
    Hi,
    i guess all the old, common iron-pigs are big endian; sun, pa-risc, power, ...

    So you mean the endianness with RS/6000 is selectable, as with IA-64 (Itanium)? https://en.wikipedia.org/wiki/IA-64

    Never had time to figure that out. Interesting :)

    Regards,
    Stefan

    Louis Ohland schrieb am Montag, 7. Februar 2022 um 15:46:17 UTC+1:
    The biggest issue will be to recognize the MCMaster. AIX does not use
    ADFs, but an ODM. The RS/6000 must be successfully IML'd, and the
    "ADFUTIL" run in order to process the MCMaster's ADF, and IIRC there is
    an Innnn.ADF as well.

    Without successful addition of the MCMaster to the ODM, the MCMaster
    will NOT be recognized and activated during set-up. All it will do is continue to dream in the endless sleep....

    Then, there is the torrid gender-bender confusion of an Intel CPU on a
    POWER system. It could be done, [68K, Transputer] other CPUs have been
    run in PS/2s, but they had smart folks writing code to enable it.

    POWER is big Endian, 486 is little Endian. IMHO, PS/2s are single image systems, biased toward intel CPUs. RS/6000s are dual image systems,
    where the MCA is accessed through a "bridge". POWER systems use markedly different IML, memory controllers, and it's all run by AIX.

    So... in short, it is Possumble, but you need some pretty heavy
    AIX-chops to get it to work.

    An aside, even the lowly -320 could probably curb-stomp a 486.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Louis Ohland@21:1/5 to schimmi on Mon Feb 7 19:37:52 2022
    Selectable? Not sure what you think I said

    POWER is big endian

    there is a very simple, very fast bit swap circuit that transposes "big
    endian" traffic to/from the MCA bus, which is little endian.

    The POWER portion is only Big Endian. MCA bus is only Little Endian.

    On 2/7/2022 16:50, schimmi wrote:
    Hi,
    i guess all the old, common iron-pigs are big endian; sun, pa-risc, power, ...

    So you mean the endianness with RS/6000 is selectable, as with IA-64 (Itanium)?
    https://en.wikipedia.org/wiki/IA-64

    Never had time to figure that out. Interesting :)

    Regards,
    Stefan

    Louis Ohland schrieb am Montag, 7. Februar 2022 um 15:46:17 UTC+1:
    The biggest issue will be to recognize the MCMaster. AIX does not use
    ADFs, but an ODM. The RS/6000 must be successfully IML'd, and the
    "ADFUTIL" run in order to process the MCMaster's ADF, and IIRC there is
    an Innnn.ADF as well.

    Without successful addition of the MCMaster to the ODM, the MCMaster
    will NOT be recognized and activated during set-up. All it will do is
    continue to dream in the endless sleep....

    Then, there is the torrid gender-bender confusion of an Intel CPU on a
    POWER system. It could be done, [68K, Transputer] other CPUs have been
    run in PS/2s, but they had smart folks writing code to enable it.

    POWER is big Endian, 486 is little Endian. IMHO, PS/2s are single image
    systems, biased toward intel CPUs. RS/6000s are dual image systems,
    where the MCA is accessed through a "bridge". POWER systems use markedly
    different IML, memory controllers, and it's all run by AIX.

    So... in short, it is Possumble, but you need some pretty heavy
    AIX-chops to get it to work.

    An aside, even the lowly -320 could probably curb-stomp a 486.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Louis Ohland@21:1/5 to Louis Ohland on Mon Feb 7 22:04:30 2022
    Upon further thought, I wonder how a PS/2's support chipset differs from
    a POWER system's support chipset?

    PS/2 DMA, arbitration, memory compared to POWER DMA, arbitration, memory

    Methinks the bus itself MIGHT be endian agnostic, but the support chips
    and I/O might beg to differ... Dunno.

    On 2/7/2022 19:37, Louis Ohland wrote:
    Selectable? Not sure what you think I said

    POWER is big endian

    there is a very simple, very fast bit swap circuit that transposes "big endian" traffic to/from the MCA bus, which is little endian.

    The POWER portion is only Big Endian. MCA bus is only Little Endian.

    On 2/7/2022 16:50, schimmi wrote:
    Hi,
    i guess all the old, common iron-pigs are big endian; sun, pa-risc,
    power, ...

    So you mean the endianness with RS/6000 is selectable, as with IA-64
    (Itanium)?
    https://en.wikipedia.org/wiki/IA-64

    Never had time to figure that out. Interesting :)

    Regards,
    Stefan

    Louis Ohland schrieb am Montag, 7. Februar 2022 um 15:46:17 UTC+1:
    The biggest issue will be to recognize the MCMaster. AIX does not use
    ADFs, but an ODM. The RS/6000 must be successfully IML'd, and the
    "ADFUTIL" run in order to process the MCMaster's ADF, and IIRC there is
    an Innnn.ADF as well.

    Without successful addition of the MCMaster to the ODM, the MCMaster
    will NOT be recognized and activated during set-up. All it will do is
    continue to dream in the endless sleep....

    Then, there is the torrid gender-bender confusion of an Intel CPU on a
    POWER system. It could be done, [68K, Transputer] other CPUs have been
    run in PS/2s, but they had smart folks writing code to enable it.

    POWER is big Endian, 486 is little Endian. IMHO, PS/2s are single image
    systems, biased toward intel CPUs. RS/6000s are dual image systems,
    where the MCA is accessed through a "bridge". POWER systems use markedly >>> different IML, memory controllers, and it's all run by AIX.

    So... in short, it is Possumble, but you need some pretty heavy
    AIX-chops to get it to work.

    An aside, even the lowly -320 could probably curb-stomp a 486.


    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Grant Taylor@21:1/5 to Louis Ohland on Tue Feb 8 23:18:40 2022
    On 2/7/22 6:37 PM, Louis Ohland wrote:
    Selectable? Not sure what you think I said

    POWER is big endian

    I'm not so sure about that.

    If you talk about POWER as in the RS/6000, sure.

    If you talk about POWER as in the contemporary POWER 8, then not so much.

    My understanding is that contemporary POWER 8 (or there about) is
    bi-endian. Meaning that code can select which endian is run, BIG or
    little. It can also be mixed such that some processes are BIG and other processes are little.

    The POWER portion is only Big Endian.

    I question the veracity of that statement.



    --
    Grant. . . .
    unix || die

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Louis Ohland@21:1/5 to Grant Taylor on Wed Feb 9 07:58:35 2022
    The origin of the Endian-ness was a MCMaster on RS/6000.

    Micro Channel.

    The POWER 8 stuff is interesting, but pounding an MCMaster into it will
    leave some scratches...

    On 2/9/2022 00:18, Grant Taylor wrote:
    On 2/7/22 6:37 PM, Louis Ohland wrote:
    Selectable? Not sure what you think I said

    POWER is big endian

    I'm not so sure about that.

    If you talk about POWER as in the RS/6000, sure.

    If you talk about POWER as in the contemporary POWER 8, then not so much.

    My understanding is that contemporary POWER 8 (or there about) is bi-endian.  Meaning that code can select which endian is run, BIG or little.  It can also be mixed such that some processes are BIG and other processes are little.

    The POWER portion is only Big Endian.

    I question the veracity of that statement.




    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Grant Taylor@21:1/5 to schimmi on Fri Feb 11 17:07:20 2022
    On 2/11/22 4:33 PM, schimmi wrote:
    i apologize if I failed to capture the situation. With selctable I've
    meant the ability switching between big and little endian, as Itanium
    can do.

    I'm not familiar with Itanium so have no bases for comparison.

    I don't know much about contemporary POWER other than it is possible for
    them to run bi-endian code at the same time. E.g. one process is big
    endian and another process is little endian. (At least that's my recollection.)



    --
    Grant. . . .
    unix || die

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From schimmi@21:1/5 to Louis Ohland on Fri Feb 11 15:33:29 2022
    Louis Ohland schrieb am Mittwoch, 9. Februar 2022 um 15:00:53 UTC+1:
    The origin of the Endian-ness was a MCMaster on RS/6000.

    Micro Channel.

    The POWER 8 stuff is interesting, but pounding an MCMaster into it will
    leave some scratches...
    On 2/9/2022 00:18, Grant Taylor wrote:
    On 2/7/22 6:37 PM, Louis Ohland wrote:
    Selectable? Not sure what you think I said

    POWER is big endian

    I'm not so sure about that.

    If you talk about POWER as in the RS/6000, sure.

    If you talk about POWER as in the contemporary POWER 8, then not so much.

    My understanding is that contemporary POWER 8 (or there about) is bi-endian. Meaning that code can select which endian is run, BIG or little. It can also be mixed such that some processes are BIG and other processes are little.

    The POWER portion is only Big Endian.

    I question the veracity of that statement.



    Hey,
    i apologize if I failed to capture the situation. With selctable I've meant the ability switching between big and little endian, as Itanium can do.
    Regards,
    Stefan

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)