Re: Modern disk system for our -11s - DEC

This is a discussion on Re: Modern disk system for our -11s - DEC ; toby@telegraphics.com.au (Toby Thain) wrote in message news: ... > "McCollum, Peter" wrote in message news: ... > > ... > > > > - Given that somebody must build the basic I/F HW, the real "work" of > > the ...

+ Reply to Thread
Results 1 to 11 of 11

Thread: Re: Modern disk system for our -11s

  1. Re: Modern disk system for our -11s

    toby@telegraphics.com.au (Toby Thain) wrote in message news:...
    > "McCollum, Peter" wrote in message news:<9B832BEB407A774AA0520CCFC232219704670823@CXOEXC11. AMERICAS.CPQCORP.NET>...
    > > ...
    > >
    > > - Given that somebody must build the basic I/F HW, the real "work" of
    > > the project is in the emulation SW. The RXV11 code is no more than a few
    > > hours to implement. The RXV12 is only slightly more, given that DMA
    > > functionality is in place. MSCP is a *lot* more work, based on what I've
    > > skimmed in the MSCP Spec.

    >
    > This is the part I am working on. I have studied the docs and simh's
    > implementation (and have skimmed some class drivers) and I believe
    > it's very do-able.


    For reference, I just stumbled across another old topical thread. This
    post shows that someone was thinking along very similar lines back in
    '02:
    http://mail-index.netbsd.org/port-va...1/24/0008.html
    (thread starts http://mail-index.netbsd.org/port-va...1/21/0006.html

    Emanuel Stiebler remarked that it's a perennial winter-time topic:
    http://mail-index.netbsd.org/port-va...1/02/0017.html )

    That's about the same time Chuck Dickman announced his adapter:
    http://mail-index.netbsd.org/port-va...1/01/0010.html

    This post mentions RSX sources to its MSCP driver (
    http://mail-index.netbsd.org/port-va...1/24/0009.html ). I have
    most of the other references mentioned (simh, UDA50 kit), but I do not
    have this driver. A VMS class driver seemed to be included with the
    patent doc but it's not machine readable and barely human readable.
    Can anyone help privately with RSX/RT/other DEC driver code for
    reference purposes?

    --Toby

    (crossposted for better effect)

    >
    > --Toby
    >
    > >
    > > ...
    > >
    > > Pete


  2. Re: Modern disk system for our -11s

    toby@telegraphics.com.au (Toby Thain) wrote in message news:...
    > ...
    > Emanuel Stiebler remarked that it's a perennial winter-time topic:
    > http://mail-index.netbsd.org/port-va...1/02/0017.html )


    And he should know, he started a similar thread in 1998:
    http://mail-index.netbsd.org/port-va...2/02/0010.html

    To which John Wilson responded
    "Sounds like a cool project. Just so you know, the SCSI side is the
    easiest (since you can use your favorite SCSI chip to do all the
    work), being a Q-bus bus master is a bit more complicated, and writing
    the firmware to do MSCP/TMSCP is quite a project."
    http://mail-index.netbsd.org/port-va...2/02/0015.html

    Unfortunately the thread dissolved into the usual SCSI-IDE debate...
    http://mail-index.netbsd.org/port-va...2/04/0032.html
    vs
    http://mail-index.netbsd.org/port-va...2/05/0005.html

    And the inevitable suggestions of bridging buses:
    http://mail-index.netbsd.org/port-va...2/06/0006.html

    I'm just collecting these links here so they're not forgotten
    The sad part is, after all these discussions over the years, nobody
    has yet delivered a hobbyist controller... At least we may have a
    three-horse race now, in Tobias Russell, vax9000, and myself.

    --Toby

  3. Re: Modern disk system for our -11s

    Toby Thain wrote:

    > toby@telegraphics.com.au (Toby Thain) wrote in message
    > news:...
    >> ...
    >> Emanuel Stiebler remarked that it's a perennial winter-time topic:
    >> http://mail-index.netbsd.org/port-va...1/02/0017.html )

    >
    > And he should know, he started a similar thread in 1998:
    > http://mail-index.netbsd.org/port-va...2/02/0010.html
    >
    > To which John Wilson responded
    > "Sounds like a cool project. Just so you know, the SCSI side is the
    > easiest (since you can use your favorite SCSI chip to do all the
    > work), being a Q-bus bus master is a bit more complicated, and writing
    > the firmware to do MSCP/TMSCP is quite a project."
    > http://mail-index.netbsd.org/port-va...2/02/0015.html
    >
    > Unfortunately the thread dissolved into the usual SCSI-IDE debate...
    > http://mail-index.netbsd.org/port-va...2/04/0032.html
    > vs
    > http://mail-index.netbsd.org/port-va...2/05/0005.html
    >
    > And the inevitable suggestions of bridging buses:
    > http://mail-index.netbsd.org/port-va...2/06/0006.html
    >
    > I'm just collecting these links here so they're not forgotten
    > The sad part is, after all these discussions over the years, nobody
    > has yet delivered a hobbyist controller... At least we may have a
    > three-horse race now, in Tobias Russell, vax9000, and myself.

    I might be the first to get there. Everything is ready except the
    block DMA bus master state machine. I spent much time to make it perfect,
    otherwise I am already there.

    vax, 9000

    >
    > --Toby



  4. Re: Modern disk system for our -11s



    Toby Thain wrote:

    > toby@telegraphics.com.au (Toby Thain) wrote in message news:...
    >
    >>...
    >>Emanuel Stiebler remarked that it's a perennial winter-time topic:
    >>http://mail-index.netbsd.org/port-va...1/02/0017.html )

    >
    >
    > And he should know, he started a similar thread in 1998:
    > http://mail-index.netbsd.org/port-va...2/02/0010.html


    In trying to answer his question "What makes this (SCSI) thing so
    expensive?..." ....
    The problem with SCSI is that DEC didn't really get on that bandwagon
    soon enough and therefore didn't put SCSI boot support into the boot
    PROMs available for most of the earlier Unibus and Q-bus systems. In
    fact, what most third party mfgrs. did was to put a microcoded bit slice
    engine on their SCSI disk controller, and have the microcode emulate the
    MSCP protocol interface. You ended up with a real bastard of a system:
    the Unibus/QBus end looked like the MSCP register set while
    the disk drive end was pure (or as pure as it could be) SCSI. Some
    controllers even had support for SCSI tape drives and used the microcode
    to emulate the TMSCP protocol (or, in some cases, the TM11/TU80/etc.
    controller registers).
    (Those not aware of history are doomed to repeat it...)

    >
    > To which John Wilson responded
    > "Sounds like a cool project. Just so you know, the SCSI side is the
    > easiest (since you can use your favorite SCSI chip to do all the
    > work), being a Q-bus bus master is a bit more complicated, and writing
    > the firmware to do MSCP/TMSCP is quite a project."
    > http://mail-index.netbsd.org/port-va...2/02/0015.html
    >
    > Unfortunately the thread dissolved into the usual SCSI-IDE debate...
    > http://mail-index.netbsd.org/port-va...2/04/0032.html
    > vs
    > http://mail-index.netbsd.org/port-va...2/05/0005.html
    >
    > And the inevitable suggestions of bridging buses:
    > http://mail-index.netbsd.org/port-va...2/06/0006.html
    >
    > I'm just collecting these links here so they're not forgotten
    > The sad part is, after all these discussions over the years, nobody
    > has yet delivered a hobbyist controller... At least we may have a
    > three-horse race now, in Tobias Russell, vax9000, and myself.
    >
    > --Toby


  5. Re: Modern disk system for our -11s

    John Hudak wrote:

    >> And he should know, he started a similar thread in 1998:
    >> http://mail-index.netbsd.org/port-va...2/02/0010.html

    >
    > In trying to answer his question "What makes this (SCSI) thing so
    > expensive?..." ....


    Great thing, this internet. You write a question in 1998, and get an
    answer in 2004 ;-)



  6. Re: Modern disk system for our -11s

    E.S. wrote:

    > John Hudak wrote:
    >
    >>> And he should know, he started a similar thread in 1998:
    >>> http://mail-index.netbsd.org/port-va...2/02/0010.html

    >>
    >> In trying to answer his question "What makes this (SCSI) thing so
    >> expensive?..." ....

    >
    > Great thing, this internet. You write a question in 1998, and get an
    > answer in 2004 ;-)

    This is truely amazing, though that question was answered long ago.

    vax, 9000

  7. Re: Modern disk system for our -11s

    "vax, 9000" wrote in message news:...
    > ...
    > I might be the first to get there. Everything is ready except the
    > block DMA bus master state machine. I spent much time to make it perfect,
    > otherwise I am already there.


    We are only beginning, although our design is rapidly taking shape and
    reference materials and test hardware are coming together. I look
    forward to seeing your published work (I'm familiar with your
    geocities site; if you want "real" domain hosting for your project,
    please contact me privately). It sounds like you will comfortably meet
    your self-imposed deadline.

    --Toby

    >
    > vax, 9000
    >
    > >
    > > --Toby


  8. Re: Modern disk system for our -11s


    "John Hudak" wrote in message
    news:416D3198.3050100@sei.cmu.edu...
    >
    >
    > Toby Thain wrote:
    >
    > > toby@telegraphics.com.au (Toby Thain) wrote in message

    news:...
    > >
    > >>...
    > >>Emanuel Stiebler remarked that it's a perennial winter-time topic:
    > >>http://mail-index.netbsd.org/port-va...1/02/0017.html )

    > >
    > >
    > > And he should know, he started a similar thread in 1998:
    > > http://mail-index.netbsd.org/port-va...2/02/0010.html

    >
    > In trying to answer his question "What makes this (SCSI) thing so
    > expensive?..." ....
    > The problem with SCSI is that DEC didn't really get on that bandwagon
    > soon enough and therefore didn't put SCSI boot support into the boot
    > PROMs available for most of the earlier Unibus and Q-bus systems. In
    > fact, what most third party mfgrs. did was to put a microcoded bit slice
    > engine on their SCSI disk controller, and have the microcode emulate the
    > MSCP protocol interface. You ended up with a real bastard of a system:
    > the Unibus/QBus end looked like the MSCP register set while
    > the disk drive end was pure (or as pure as it could be) SCSI. Some
    > controllers even had support for SCSI tape drives and used the microcode
    > to emulate the TMSCP protocol (or, in some cases, the TM11/TU80/etc.
    > controller registers).
    > (Those not aware of history are doomed to repeat it...)
    >


    A la the CMD CQD controllers and I think DEC did a similar product. IIRC
    they worked very well.

    Surely there must be plenty of the CMD cards on the second hand market these
    day, why reinvent the wheel ?

    Dave



  9. Re: Modern disk system for our -11s

    DaveG schrieb:

    >
    >
    > A la the CMD CQD controllers and I think DEC did a similar product. IIRC
    > they worked very well.
    >
    > Surely there must be plenty of the CMD cards on the second hand market these
    > day, why reinvent the wheel ?
    >


    In fact the demand is much higher than the suppley...
    what happens if one comes around can be seen here:
    http://cgi.ebay.com/ws/eBayISAPI.dll...tem=5131612327 or
    http://cgi.ebay.com/ws/eBayISAPI.dll...tem=5131808150
    The CMD was the first Unibus device I saw in the last 2 years....

    Second hand dealers ask typicalley 400 - 600 $$ for such modules. There
    are just too many vaxen still in productive use that require replacement
    of difficult to maintain older disc technologie. This keeps prices of
    SCSI-controllers out of hobby reach.

    Reinventing the wheel sounds silly, but most of the custom
    semiconductors used in those designs of the mid 80's arn't available
    anymore, so it may really be required.

    Frank

  10. Re: Modern disk system for our -11s

    John Hudak wrote in message news:<416D3198.3050100@sei.cmu.edu>...
    > The problem with SCSI is that DEC didn't really get on that bandwagon
    > soon enough and therefore didn't put SCSI boot support into the boot
    > PROMs available for most of the earlier Unibus and Q-bus systems.


    With a Q-bus RQZX1 (the DEC SCSI controller for the Q-bus) there was
    no problem booting with the DU driver.

    > In
    > fact, what most third party mfgrs. did was to put a microcoded bit slice
    > engine on their SCSI disk controller


    Well, the DEC UDA50/KDA50 was microcoded bit slice. But most 3rd-party
    MSCP controllers were just conventional microcontrollers. (And then
    there were some Websters, where the bit-slice CPU actually ran 6502
    code!)

    > and have the microcode emulate the
    > MSCP protocol interface. You ended up with a real bastard of a system


    The MSCP protocol was *intended* to hide the actual interface and
    geometry from the OS. If anything it was a little over-general for
    use on a PDP-11 (but it scaled all the way up to VAXclusters, with
    elaborate port/class drivers that the PDP-11 world didn't fully appreciate...
    although there were many big PDP-11 sites that used dual porting to good
    advantage.)

    Tim.

  11. Re: Modern disk system for our -11s

    "DaveG" wrote in message news:...
    > "John Hudak" wrote in message
    > news:416D3198.3050100@sei.cmu.edu...
    > >
    > >
    > > Toby Thain wrote:
    > >
    > > > toby@telegraphics.com.au (Toby Thain) wrote in message

    > news:...
    > > >
    > > >>...
    > > >>Emanuel Stiebler remarked that it's a perennial winter-time topic:
    > > >>http://mail-index.netbsd.org/port-va...1/02/0017.html )
    > > >
    > > >
    > > > And he should know, he started a similar thread in 1998:
    > > > http://mail-index.netbsd.org/port-va...2/02/0010.html

    > >
    > > In trying to answer his question "What makes this (SCSI) thing so
    > > expensive?..." ....
    > > [snip]
    > >

    >
    > A la the CMD CQD controllers and I think DEC did a similar product. IIRC
    > they worked very well.


    There were quite a number of brands at one time (I can think of
    eight).

    >
    > Surely there must be plenty of the CMD cards on the second hand market these
    > day, why reinvent the wheel ?


    While they do appear on ebay occasionally, they are worth a lot more
    on that market than hobbyists can typically afford. Scarcity (as
    always) drives the prices, so this is unlikely to change soon. The
    fact that people are still buying them for commercial use helps
    inflate prices. Dealers typically quote around five times the ebay
    price (i.e. four figures). This chagrins people like myself who have
    more machines than working disks (and the disks that *do* work are
    bound to die eventually). So it's hardly a frivolous project, as the
    perennial threads begging for such a card seem to indicate...

    --Toby

    >
    > Dave


+ Reply to Thread