X-Mode

[Edward Feustel]

I am interested in obtaining an X-Mode Manual and Prime tools for
emulated X-Mode on Primos.

Thanks,
Ed Feustel

[Scott Dorsey]

In article <scdcuhd5fj0nd151j...@4ax.com>,

Edward Feustel <efeu...@feustel.us> wrote:
>I am interested in obtaining an X-Mode Manual and Prime tools for
>emulated X-Mode on Primos.

I sent a complete set of Primos 16 manuals to Bitsavers and most of a
set of Primos 19 manuals, probably a decade or so ago. If they don't have
them online yet, ask them.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Dennis Boone]

> >I am interested in obtaining an X-Mode Manual and Prime tools for
> >emulated X-Mode on Primos.

> I sent a complete set of Primos 16 manuals to Bitsavers and most of a
> set of Primos 19 manuals, probably a decade or so ago. If they don't have
> them online yet, ask them.

Pretty much everything Al has posted, plus more scanned by others,
is also on https://sysovl.info/reference_prime.html

But I don't think much X-mode material was in the published architecture
manuals.

De

[ukctya]

Those PETs you scanned are a really fascinating look inside an engineering organization

The proposal to write a brand new I/O system from scratch I think did not go anywhere, did it? Sounds a bit high risk!

What's this NSP project they keep mentioning?

I wish there were more, particularly from later in the 80s!

[Edward Feustel]

Thanks De.
Unfortunately X-Mode did not quite make into a product. When I was
laid off in 1992 they were working on a CMOS CPU that had the
microcode for X-Mode including IEEE floating point. So the latest
instruction manuals were 1987. What I am looking for are internal
manuals by the compiler people or the OS people.

I had put my latest stuff on EXL tapes and this info was lost.
My older stuff was on Primos tapes and I got all of it off.

Best regards,
Ed

[Dennis Boone]

Ed,

> Unfortunately X-Mode did not quite make into a product. When I was
> laid off in 1992 they were working on a CMOS CPU that had the
> microcode for X-Mode including IEEE floating point. So the latest
> instruction manuals were 1987. What I am looking for are internal
> manuals by the compiler people or the OS people.

That's about what I was remembering. I assume there were PETs for the
architectural changes, and for software support of same?

The latest scans I have are 1987, and I don't think I have any paper
that's newer, though that could be buried in a stack somewhere easily
enough. But there's at least a DOC9473-3LA that's later than that 1987
Rev.21 edition.

I would dearly love it if Dons Slutz or Koch (or any other engineers, I
just pick on them because I know that as of fairly recently they were
still around in MA, and one of them had an annual bbq) had paper or
media in their possession and could share it to the preservation
community.

IIRC your paper materials went to a university collection. Rice?
Babbaage?

> I had put my latest stuff on EXL tapes and this info was lost.
> My older stuff was on Primos tapes and I got all of it off.

Are the tapes destroyed, or you just don't have a mechanism for reading
them? If the latter, I think we can find a solution.

I should probbably mention that there are two 50-Series emulators, both
on github, in case you weren't aware.

De

[Dennis Boone]

> Those PETs you scanned are a really fascinating look inside an
> engineering organization

Internal documentation is always interesting.

I should point out that I presonally scanned relatively little of
the stuff hosted on sysovl. The index pages show who did the work.

> The proposal to write a brand new I/O system from scratch I think did
> not go anywhere, did it? Sounds a bit high risk!

Not just software. One of Prime's major problems by the mid-80s was the
I/O backplane maxing out at around 5 MB/s. No software solution was
going to fix that. That was beginning to be an issue with SMD-E drives
in the 80s, never mind the SCSI drives in the 5000 series machines.

> What's this NSP project they keep mentioning?

It looks to me like it might be a prior project name for X mode.

> I wish there were more, particularly from later in the 80s!

Amen!

De

[Dennis Boone]

> The proposal to write a brand new I/O system from scratch I think did
> not go anywhere, did it? Sounds a bit high risk!

Oh, and I neglected to answer the actual question. A real new I/O
system never shipped. They had the familiar problem that fixing the
limitations would have meant discarding all compatibility with older
controllers, and probably a lot of software as well. Prime never liked
to break compatibility! But you can see more push to intelligent
controllers for various things (comms, disk). The first MPC-based stuff
appeared pretty early, but in the mid-late 80s, they did the 80286-based
LAN Host Controller for ethernet-based networking and terminal service,
the assorted intelligent disk controllers, three rounds of intelligent
communications (serial for terminals and network links) controllers,
etc.

But perhaps the highest risk was _not_ doing something. One's lunch
will surely get eaten by others if one doesn't solve one's limitations.
Microcomputers were already eating their lunch.

De

[bill]

Why does your name look so familiar? Did you attend the Orlando
PUG Conference back in the day?

bill

[Edward Feustel]

De,
I sent almost all my non-confidential material to the Computer History
Museum on the west coast. I think they were all copied to bitsavers
and are available there.

I sent my tanberg tape reader and three copies of my prime exl tapes
to Don Koch who was unable to access the material. He may still have
the tapes. If you had access to an EXL with the tapes, I would see
if he still has the tapes. There is a good deal of material I would
like to retrieve from them. I wish I had copied all of it over to
the Primos Side. Those archives all survived 100%.

Ed

[Edward Feustel]

Bill,
I am not certain where we might have met.
I started at Prime in December 1980 as a Senior Research Consultant
under Bob Gordon. I spent time in OS working on Prime Unix.
I spent time under Len Halio working on debugging EXL terminal I/O
as a Senior Technicl Consultant in Milford
I worked for Walter Jones first in CPU and then as a Principal
Technical Consultant on his staff when he was VP of Engineering.
I was laid off in 1992.
I went to work for IDA in Alexandria VA and spent an inordinate amount
of time in Orlando working on simulation architecture. Perhaps I met
you there post=Prime.

Ed


On Sat, 11 Feb 2023 19:05:37 -0500, bill <bill.gu...@gmail.com>
wrote:

[Dennis Boone]

> I sent my tanberg tape reader and three copies of my prime exl tapes
> to Don Koch who was unable to access the material. He may still have
> the tapes. If you had access to an EXL with the tapes, I would see
> if he still has the tapes. There is a good deal of material I would
> like to retrieve from them. I wish I had copied all of it over to
> the Primos Side. Those archives all survived 100%.

Ed,

A couple of thoughts:

1. I'm pretty sure there's no need for an actual EXL to read the tapes.
There may be some work needed with the raw images to sort out a blocking
scheme or similar. The 50-Series cart drive controllers wrapped
arbitrary sized host data blocks to store them on the QIC 512-byte fixed
blocks, for example. But data massaging of that sort usually isn't _too_
hard.

2. Chuck Guzis at Sydex has years of expertise at reading all sorts of
media. It would be worth talking to him about whether he thinks he
could read the tapes.

https://www.sydex.com/

De

[Jim Wilcoxson]

On Wednesday, February 15, 2023 at 12:43:44 PM UTC-5, Dennis Boone wrote:
> > I sent my tanberg tape reader and three copies of my prime exl tapes
> > to Don Koch who was unable to access the material. He may still have
> > the tapes. If you had access to an EXL with the tapes, I would see
> > if he still has the tapes. There is a good deal of material I would
> > like to retrieve from them. I wish I had copied all of it over to
> > the Primos Side. Those archives all survived 100%.
> Ed,
>
> A couple of thoughts:
>
> 1. I'm pretty sure there's no need for an actual EXL to read the tapes.
> There may be some work needed with the raw images to sort out a blocking
> scheme or similar. The 50-Series cart drive controllers wrapped
> arbitrary sized host data blocks to store them on the QIC 512-byte fixed
> blocks, for example. But data massaging of that sort usually isn't _too_
> hard.

I've done this before with weird tapes and it's not too hard. Someone sent me some tapes from a power company that were on weird cartridges. Dennis is right that it's a 2-step process: read raw tape blocks, figure out what kind of blocking and headers have been added, strip all that out, and for Magsav tapes, run it through the emulator version of magrst, which doesn't care about physical tape records. I'm assuming the EXL wrote tar tapes, so you might have to grab a copy of tar and hack it a bit. Or if the EXL wrote some kind of UNIXy Magsav tape, you might have to hack the emulator version of magrst a bit.

I no longer have any tape gear, but if someone can read the tapes with mtread (an emulator utility program that reads a raw tape), I will help decipher it.

Jim

[bill]

On 2/15/2023 8:34 AM, Edward Feustel wrote:
> Bill,
> I am not certain where we might have met.
> I started at Prime in December 1980 as a Senior Research Consultant
> under Bob Gordon.

I worked with Prime 50 Series at USMA/West Point. First while still
in the Army and then for a short while with Martin Marietta as a DOD
contractor. Sadly, my work with Primes ended around 1985 except for
the 2250 I had at home. :-)

> I spent time in OS working on Prime Unix.

We beta-tested Primix but I was most interested in the work that
Bill Lenhart did up in NH. Sad that it never got released into
the wild.

> I spent time under Len Halio working on debugging EXL terminal I/O
> as a Senior Technicl Consultant in Milford
> I worked for Walter Jones first in CPU and then as a Principal
> Technical Consultant on his staff when he was VP of Engineering.
> I was laid off in 1992.
> I went to work for IDA in Alexandria VA and spent an inordinate amount
> of time in Orlando working on simulation architecture. Perhaps I met
> you there post=Prime.

My last trip to Orlando (not Prime related) was to the Martin Marietta
facility around 85-86. By 88 I was out of the government work and into'
academia where I spent the rest of my career (other than my military
time as I returned to part-time service in the Guard and Reserves as
an IT Warrant Officer).

Oh well, maybe we never met and I only know you from here. Mind isn't
as clear as it used to be. :-)

bill

[Edward Feustel]

Jim,
The EXL tapes are cartridge tapes. The tape units had a modification
that allowed a blocksize larger than standard cartridge tapes. I had
bought a tandberg tape unit with a modified block size. But it had
a bad loading problem that tended to destroy the tapes (as reported
by Don. It would have to be an EXL cartridge unit or one that could
read 1040 byte records (as I remember the byte size).

Thanks for your work on the emulators which obviously don't do
X-Mode.

Ed Feustel

[Edward Feustel]

Perhaps it was related to Software Tools, the programming language
Icon, or Guy Almes.

Ed Feustel


On Thu, 16 Feb 2023 08:28:14 -0500, bill <bill.gu...@gmail.com>
wrote:

[ukctya]

On Friday, February 17, 2023 at 5:51:22 AM UTC-5, Edward Feustel wrote:
> Jim,

>
> Thanks for your work on the emulators which obviously don't do
> X-Mode.
>

I second this. The emulators are an amazing piece of work.

As for the EXL I recall back around the 1988 timeframe Prime lent us one (we were a small s/w house)

I ported some C code from a Sun 3/60 to the EXL which the Prime rep referred to as a "Matchbox"

However, they took the machine away, and nothing came of it.

[bill]

On 2/17/2023 5:55 AM, Edward Feustel wrote:
> Perhaps it was related to Software Tools, the programming language
> Icon, or Guy Almes.
>

If you were involved in STUG that was probably the place. Although
I never got to go to any of their get-togethers. I was very interested
in STVOS and still am today although the users group meetings are pretty
lonely with only me there.

I always enjoy pointing out to people that POSIX is nothing but STVOS
warmed over and served badly. Think what it could be doing today if
it hadn't been abandoned only to be restarted from the very beginning
decades later.

bill

[Dennis Boone]

> The EXL tapes are cartridge tapes. The tape units had a modification
> that allowed a blocksize larger than standard cartridge tapes. I had
> bought a tandberg tape unit with a modified block size. But it had
> a bad loading problem that tended to destroy the tapes (as reported
> by Don. It would have to be an EXL cartridge unit or one that could
> read 1040 byte records (as I remember the byte size).

Ed,

Are you _ABSOLUTELY SURE_ it was the tape units themselves, not the
controllers? Because on the 50-Series side, I'm pretty sure it was the
controllers that did the blocking magic. I've read PRIMOS distribution
tapes from QIC carts with unmodified drives, and done the deblocking of
this scheme. The 50-Series disks, after the early period, used 1040
_word_ disk sector size, with 1024 data words and 16 file system
overhead words as a header. PRIMOS, having grown up with half inch
7-track and 9-track reel tapes, was accustomed to being having variable
on-tape block sizes as large as the available Ring-0 buffer could hold.
The easiest path to getting MAG* and other software to work was, I
presume, to get the controllers to present 512 fixed to the drive and
longer variable to the host.

It seems extremely odd that a unix system would depend on such modified
drives and unusual (for unix) block size.

There are known issues with the little rubber bands in the QIC carts,
for example, that also have known workarounds. And rubber wheels in the
drive train of such mechanisms also get gooey. Some brands worse than
others, and some folks have had this sort of wheel rebuilt.

I still think it'd be worth at least talking to Chuck Guzis about these
tapes. If there's shareable 50-Series stuff on the tapes, I'd be
willing to help cover the cost.

De

[Jim Wilcoxson]

Prime was famous for convincing customers that the hardware they sold was somehow "enhanced", to prevent customers from using off-the-shelf products. For example, their 8mm drive was way more expensive than buying one from Exabyte, but only the Prime 8mm drive would work on a Prime system. However, I bought a regular 8mm drive on eBay, hooked it up to a PC SCSI controller, and could read old Prime 8mm tapes just fine.

I did a little searching and this article looked interesting. Maybe this guy would be willing to help out with reading a Prime tape. I'll be glad to share expenses too:

http://www.hp9825.com/html/qic_and_the_dead.html

Jim

[Scott Dorsey]

Edward Feustel <efeu...@feustel.us> wrote:
>Jim,
>The EXL tapes are cartridge tapes. The tape units had a modification
>that allowed a blocksize larger than standard cartridge tapes. I had
>bought a tandberg tape unit with a modified block size. But it had
>a bad loading problem that tended to destroy the tapes (as reported
>by Don. It would have to be an EXL cartridge unit or one that could
>read 1040 byte records (as I remember the byte size).

You used to have to replace the belts and the rubber idler puck on
those periodically as part of PMs. If you can't replace them, try
cleaning them with 409 and testing on a junk tape.

[Dennis Boone]

> Prime was famous for convincing customers that the hardware they sold
> was somehow "enhanced", to prevent customers from using off-the-shelf
> products. For example, their 8mm drive was way more expensive than
> buying one from Exabyte, but only the Prime 8mm drive would work on a
> Prime system. However, I bought a regular 8mm drive on eBay, hooked it
> up to a PC SCSI controller, and could read old Prime 8mm tapes just
> fine.

Right, it's one thing to prevent the drive from working with the Prime
host by having the controller refuse to accept unmodified device
firmware. There is such firmware in the Prime labeled Exabyte drives.
It's altogether another to dig into the third party device far enough to
really modify core elements of its behavior.

By the time Prime shipped the EXL stuff, there were additional
considerations. They already knew they were under pressure from
commodity microprocessor-based systems, which were limiting the market
appeal of anything proprietary, and also reduced the available
engineering budget for any given piece of gear. Also, while the EXL-3xx
line was an in-house design and might have been easier to tweak, the
7xxx systems were just rebadged MIPS systems.

De

[Edward Feustel]

Yes, I am really certain about the special code required.
I think the only stuff on the system were my e-mail as a PTC,
documents I had produced (maybe X-Mode), and various PC
oriented stuff. I wanted my e-mail.

The cartridges were 90 mb if I remember correctly.

Don Kock discovered all the difficulties of the Tandberg w.r.t.
tensioning. He was not inclined to try to get things off the tapes.
He has all three of my tapes. I don't know whether he disposed of
them. I don't know whether there are any EXLs in existance.

Best regards,
Ed

[Edward Feustel]

On Fri, 17 Feb 2023 12:39:58 -0500, bill <bill.gu...@gmail.com>
wrote:

Bill,
I can't remember ever attending a STUG. I may have provided a versions
of the Icon programming system that ran under the STUG release.
Or I suppose it could have been something I was involved with in
the Research Group.

Ed