Ultrix-11

[Bill Gunshannon]

First, let me wish all my American friends a happy Thanksgiving.
And all the best to those who don't celebrate a holiday today, too.

I have finally reached the point where I can start playing with my
PDP-11's again. One of my main interests was Ultrix-11. This is
because I have always liked Unix and this is really the only PDP-11
OS that has been officially released for hobbyist use. So.......

I have an emulated 11/73 with 4M of memory, a DEQNA, a KDA50 with 4 RA81
disks, an RQDX3 and an RXV21, and both TS11 and TK50 tapes.

I have already Y2Ked the date command (getting the date to catchup before
was a real pain.) I plan to play with the system for making install media.
I think a version on RX23 or RX33 would beat the heck out of RX50 which
is what it supports now.

The networking has some bugs that need fixing, too.

Is there any interest by others in playing with and updating Ultrix-11?
Any particular things that you think need looking at? Is there even
anyone else still running it?

And, does anyone here know what the major mistake in the final system as it
was distributed by DEC was? :-) Think "uname -a". :-)

bill


--
Bill Gunshannon | de-moc-ra-cy (di mok' ra see) n. Three wolves
bill...@cs.scranton.edu | and a sheep voting on what's for dinner.
University of Scranton |
Scranton, Pennsylvania | #include <std.disclaimer.h>

[Johnny Billquist]

On 2012-11-22 15:18, Bill Gunshannon wrote:
> First, let me wish all my American friends a happy Thanksgiving.
> And all the best to those who don't celebrate a holiday today, too.
>
> I have finally reached the point where I can start playing with my
> PDP-11's again. One of my main interests was Ultrix-11. This is
> because I have always liked Unix and this is really the only PDP-11
> OS that has been officially released for hobbyist use. So.......

Whoa! What about 2bsd?

> I have an emulated 11/73 with 4M of memory, a DEQNA, a KDA50 with 4 RA81
> disks, an RQDX3 and an RXV21, and both TS11 and TK50 tapes.

Nice setup.

> I have already Y2Ked the date command (getting the date to catchup before
> was a real pain.) I plan to play with the system for making install media.
> I think a version on RX23 or RX33 would beat the heck out of RX50 which
> is what it supports now.
>
> The networking has some bugs that need fixing, too.
>
> Is there any interest by others in playing with and updating Ultrix-11?
> Any particular things that you think need looking at? Is there even
> anyone else still running it?

Nah. Running 2bsd when I want to play Unix. :-) It could be nice to get
a copy just to check, though.

> And, does anyone here know what the major mistake in the final system as it
> was distributed by DEC was? :-) Think "uname -a". :-)

Does it just claim "Ultrix"? :-)

Johnny

[Bill Gunshannon]

In article <k8m7fp$vf2$1...@iltempo.update.uu.se>,

Johnny Billquist <b...@softjar.se> writes:
> On 2012-11-22 15:18, Bill Gunshannon wrote:
>> First, let me wish all my American friends a happy Thanksgiving.
>> And all the best to those who don't celebrate a holiday today, too.
>>
>> I have finally reached the point where I can start playing with my
>> PDP-11's again. One of my main interests was Ultrix-11. This is
>> because I have always liked Unix and this is really the only PDP-11
>> OS that has been officially released for hobbyist use. So.......
>
> Whoa! What about 2bsd?

Got that around here somewhere, too. But, like RSTS I always liked
Ultrix-11.

>
>> I have an emulated 11/73 with 4M of memory, a DEQNA, a KDA50 with 4 RA81
>> disks, an RQDX3 and an RXV21, and both TS11 and TK50 tapes.
>
> Nice setup.

Yeah, sometimes you can actually do a lot more in emulation than you
could ever hope for in real hardware.

>
>> I have already Y2Ked the date command (getting the date to catchup before
>> was a real pain.) I plan to play with the system for making install media.
>> I think a version on RX23 or RX33 would beat the heck out of RX50 which
>> is what it supports now.
>>
>> The networking has some bugs that need fixing, too.
>>
>> Is there any interest by others in playing with and updating Ultrix-11?
>> Any particular things that you think need looking at? Is there even
>> anyone else still running it?
>
> Nah. Running 2bsd when I want to play Unix. :-) It could be nice to get
> a copy just to check, though.

I'll let you kow when I have a reasonable install media ready. I am
thinkig I will make the new install media with all the sources and
everything already in place.

>
>> And, does anyone here know what the major mistake in the final system as it
>> was distributed by DEC was? :-) Think "uname -a". :-)
>
> Does it just claim "Ultrix"? :-)

Actually, when you boot it says "Ultrix-11 3.1" but "uname -a" reports
"ULTRIX-11 ultrix 3 0 PDP-11/73". I looked att he sources and somebody
forgot to change the "0" to a "1" :-).


And another interesting note. There are Digital Copyrights in all
the source files. There are no AT&T Copyrights in any of the source
files and only a few BSD Copyrights, Am I the only one who doubts
that DEC rewrote the whole operating system in a clean room environment?

[Johnny Billquist]

On 2012-11-22 23:43, Bill Gunshannon wrote:
> In article <k8m7fp$vf2$1...@iltempo.update.uu.se>,
> Johnny Billquist <b...@softjar.se> writes:
>> On 2012-11-22 15:18, Bill Gunshannon wrote:
>>> First, let me wish all my American friends a happy Thanksgiving.
>>> And all the best to those who don't celebrate a holiday today, too.
>>>
>>> I have finally reached the point where I can start playing with my
>>> PDP-11's again. One of my main interests was Ultrix-11. This is
>>> because I have always liked Unix and this is really the only PDP-11
>>> OS that has been officially released for hobbyist use. So.......
>>
>> Whoa! What about 2bsd?
>
> Got that around here somewhere, too. But, like RSTS I always liked
> Ultrix-11.

Yes, but 2bsd is also free. Which is why I reacted to your claim about
Ultrix-11 being the only thing around.

>>> I have an emulated 11/73 with 4M of memory, a DEQNA, a KDA50 with 4 RA81
>>> disks, an RQDX3 and an RXV21, and both TS11 and TK50 tapes.
>>
>> Nice setup.
>
> Yeah, sometimes you can actually do a lot more in emulation than you
> could ever hope for in real hardware.

Mostly speed, though.
Question, though. Why not emulate an 11/93?

>>> And, does anyone here know what the major mistake in the final system as it
>>> was distributed by DEC was? :-) Think "uname -a". :-)
>>
>> Does it just claim "Ultrix"? :-)
>
> Actually, when you boot it says "Ultrix-11 3.1" but "uname -a" reports
> "ULTRIX-11 ultrix 3 0 PDP-11/73". I looked att he sources and somebody
> forgot to change the "0" to a "1" :-).

Ouch. :-)

> And another interesting note. There are Digital Copyrights in all
> the source files. There are no AT&T Copyrights in any of the source
> files and only a few BSD Copyrights, Am I the only one who doubts
> that DEC rewrote the whole operating system in a clean room environment?

People were a bit more sloppy back then.

Johnny

--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: b...@softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol

[Bill Gunshannon]

In article <k8mnoc$587$1...@iltempo.update.uu.se>,

Johnny Billquist <b...@softjar.se> writes:
> On 2012-11-22 23:43, Bill Gunshannon wrote:
>> In article <k8m7fp$vf2$1...@iltempo.update.uu.se>,
>> Johnny Billquist <b...@softjar.se> writes:
>>> On 2012-11-22 15:18, Bill Gunshannon wrote:
>>>> First, let me wish all my American friends a happy Thanksgiving.
>>>> And all the best to those who don't celebrate a holiday today, too.
>>>>
>>>> I have finally reached the point where I can start playing with my
>>>> PDP-11's again. One of my main interests was Ultrix-11. This is
>>>> because I have always liked Unix and this is really the only PDP-11
>>>> OS that has been officially released for hobbyist use. So.......
>>>
>>> Whoa! What about 2bsd?
>>
>> Got that around here somewhere, too. But, like RSTS I always liked
>> Ultrix-11.
>
> Yes, but 2bsd is also free. Which is why I reacted to your claim about
> Ultrix-11 being the only thing around.

Oh, sorry. I meant DEC OSes.

>
>>>> I have an emulated 11/73 with 4M of memory, a DEQNA, a KDA50 with 4 RA81
>>>> disks, an RQDX3 and an RXV21, and both TS11 and TK50 tapes.
>>>
>>> Nice setup.
>>
>> Yeah, sometimes you can actually do a lot more in emulation than you
>> could ever hope for in real hardware.
>
> Mostly speed, though.
> Question, though. Why not emulate an 11/93?

Not in the list of supported CPU's. Can probably add the 83 and 93 later
when I have everything set up to build a newer version. Looking at
changing the partitioning scheme (at least for the RA81) as well. /usr
is way to small even for their own distribution.

But, as far as CPU goes what other than speed did the 93 really offer?

>
>>>> And, does anyone here know what the major mistake in the final system as it
>>>> was distributed by DEC was? :-) Think "uname -a". :-)
>>>
>>> Does it just claim "Ultrix"? :-)
>>
>> Actually, when you boot it says "Ultrix-11 3.1" but "uname -a" reports
>> "ULTRIX-11 ultrix 3 0 PDP-11/73". I looked att he sources and somebody
>> forgot to change the "0" to a "1" :-).
>
> Ouch. :-)

And some of the comments by Fred Canter are hilarious. :-)

>
>> And another interesting note. There are Digital Copyrights in all
>> the source files. There are no AT&T Copyrights in any of the source
>> files and only a few BSD Copyrights, Am I the only one who doubts
>> that DEC rewrote the whole operating system in a clean room environment?
>
> People were a bit more sloppy back then.

Not so sure it was sloppy. Wasn't this also a major point in the BSDI-AT&T
lawsuit? Hadn't AT&T taken all of UCB's Copyright notices out of the code
they used, like networking?

And, on one other note, I don't know if anyone ever played with it, but
the SRCKIT tape is munged as well. If you follow the instructions for
installing it some things end out int he wrong places so I doubt that
trying to build a whole system would have worked. But I think I can
fix that, too.

[Johnny Billquist]

On 2012-11-23 04:17, Bill Gunshannon wrote:
> In article <k8mnoc$587$1...@iltempo.update.uu.se>,
> Johnny Billquist <b...@softjar.se> writes:
>> On 2012-11-22 23:43, Bill Gunshannon wrote:
>>> In article <k8m7fp$vf2$1...@iltempo.update.uu.se>,
>>> Johnny Billquist <b...@softjar.se> writes:
>>>> On 2012-11-22 15:18, Bill Gunshannon wrote:
>>>>> First, let me wish all my American friends a happy Thanksgiving.
>>>>> And all the best to those who don't celebrate a holiday today, too.
>>>>>
>>>>> I have finally reached the point where I can start playing with my
>>>>> PDP-11's again. One of my main interests was Ultrix-11. This is
>>>>> because I have always liked Unix and this is really the only PDP-11
>>>>> OS that has been officially released for hobbyist use. So.......
>>>>
>>>> Whoa! What about 2bsd?
>>>
>>> Got that around here somewhere, too. But, like RSTS I always liked
>>> Ultrix-11.
>>
>> Yes, but 2bsd is also free. Which is why I reacted to your claim about
>> Ultrix-11 being the only thing around.
>
> Oh, sorry. I meant DEC OSes.

Ah. Ok, then I got it.

>>>>> I have an emulated 11/73 with 4M of memory, a DEQNA, a KDA50 with 4 RA81
>>>>> disks, an RQDX3 and an RXV21, and both TS11 and TK50 tapes.
>>>>
>>>> Nice setup.
>>>
>>> Yeah, sometimes you can actually do a lot more in emulation than you
>>> could ever hope for in real hardware.
>>
>> Mostly speed, though.
>> Question, though. Why not emulate an 11/93?
>
> Not in the list of supported CPU's. Can probably add the 83 and 93 later
> when I have everything set up to build a newer version. Looking at
> changing the partitioning scheme (at least for the RA81) as well. /usr
> is way to small even for their own distribution.

Doh. They stopped Ultrix-11 too soon. :-)

> But, as far as CPU goes what other than speed did the 93 really offer?

TOY clock. Memory and serial ports on the same card as the CPU.

>>> And another interesting note. There are Digital Copyrights in all
>>> the source files. There are no AT&T Copyrights in any of the source
>>> files and only a few BSD Copyrights, Am I the only one who doubts
>>> that DEC rewrote the whole operating system in a clean room environment?
>>
>> People were a bit more sloppy back then.
>
> Not so sure it was sloppy. Wasn't this also a major point in the BSDI-AT&T
> lawsuit? Hadn't AT&T taken all of UCB's Copyright notices out of the code
> they used, like networking?

It was, and yes. But was the removal done because they tried to hide the
origins, and pretend the code was done by them, or just because they
just wanted uniform file headers, and noone really gave much thought to
whether this was formally ok to do or not?
I think the latter, and I think that was just sloppy. But I might be wrong.

[Bill Gunshannon]

In article <k8npdd$gja$1...@iltempo.update.uu.se>,

Yes, but then, some of us think they stopped the PDP-11 too soon, :-)

>
>> But, as far as CPU goes what other than speed did the 93 really offer?
>
> TOY clock. Memory and serial ports on the same card as the CPU.

Well, the TOY clock is nice if you have real hardware, but I have
another solution. I am going to provide a utility to keep the time
synched using an old RS-232 connected GPS receiver. I havr an old
Delorme Earthmate and I figure there a thousands of them floating
around. may even look on eBay and grab a few more. The rest are no
big deal in hardware and pretty much meaningless in emulated systems.
I meant the CPU itself. And, as I said, I think speed was pretty
much it and that, too, doesn't have much meaning in an emulated
system as I figure an 11/02 and an 11/93 process instructions at the
same speed. Unless there is some throttling being done that I was
not aware of.

>
>>>> And another interesting note. There are Digital Copyrights in all
>>>> the source files. There are no AT&T Copyrights in any of the source
>>>> files and only a few BSD Copyrights, Am I the only one who doubts
>>>> that DEC rewrote the whole operating system in a clean room environment?
>>>
>>> People were a bit more sloppy back then.
>>
>> Not so sure it was sloppy. Wasn't this also a major point in the BSDI-AT&T
>> lawsuit? Hadn't AT&T taken all of UCB's Copyright notices out of the code
>> they used, like networking?
>
> It was, and yes. But was the removal done because they tried to hide the
> origins, and pretend the code was done by them, or just because they
> just wanted uniform file headers, and noone really gave much thought to
> whether this was formally ok to do or not?
> I think the latter, and I think that was just sloppy. But I might be wrong.

Well, I thought the AT&T case was that someone said "AT&T software only
has AT&T Copyrights" so the others were very deliberately removed. I
suspect it was quite the same with DEC. And people think nobody understands
copyright today!!

[Johnny Billquist]

On 2012-11-23 13:53, Bill Gunshannon wrote:
> In article <k8npdd$gja$1...@iltempo.update.uu.se>,
> Johnny Billquist <b...@softjar.se> writes:
>> On 2012-11-23 04:17, Bill Gunshannon wrote:
>>> In article <k8mnoc$587$1...@iltempo.update.uu.se>,
>>> Johnny Billquist <b...@softjar.se> writes:
>>>> On 2012-11-22 23:43, Bill Gunshannon wrote:
>>>>> In article <k8m7fp$vf2$1...@iltempo.update.uu.se>,
>>>>> Johnny Billquist <b...@softjar.se> writes:
>>>>>> On 2012-11-22 15:18, Bill Gunshannon wrote:
>>>>>>> I have an emulated 11/73 with 4M of memory, a DEQNA, a KDA50 with 4 RA81
>>>>>>> disks, an RQDX3 and an RXV21, and both TS11 and TK50 tapes.
>>>>>>
>>>>>> Nice setup.
>>>>>
>>>>> Yeah, sometimes you can actually do a lot more in emulation than you
>>>>> could ever hope for in real hardware.
>>>>
>>>> Mostly speed, though.
>>>> Question, though. Why not emulate an 11/93?
>>>
>>> Not in the list of supported CPU's. Can probably add the 83 and 93 later
>>> when I have everything set up to build a newer version. Looking at
>>> changing the partitioning scheme (at least for the RA81) as well. /usr
>>> is way to small even for their own distribution.
>>
>> Doh. They stopped Ultrix-11 too soon. :-)
>
> Yes, but then, some of us think they stopped the PDP-11 too soon, :-)

Indeed. But you can still buy a new PDP-11. Just go to Quickware
(quickware.com), unless they somehow have forgotten that they are out of
business. :-)

The software side is way murkier, unfortunately. :-(

>>> But, as far as CPU goes what other than speed did the 93 really offer?
>>
>> TOY clock. Memory and serial ports on the same card as the CPU.
>
> Well, the TOY clock is nice if you have real hardware, but I have
> another solution. I am going to provide a utility to keep the time
> synched using an old RS-232 connected GPS receiver. I havr an old
> Delorme Earthmate and I figure there a thousands of them floating
> around. may even look on eBay and grab a few more. The rest are no
> big deal in hardware and pretty much meaningless in emulated systems.
> I meant the CPU itself. And, as I said, I think speed was pretty
> much it and that, too, doesn't have much meaning in an emulated
> system as I figure an 11/02 and an 11/93 process instructions at the
> same speed. Unless there is some throttling being done that I was
> not aware of.

Right. From a CPU point itself, there is no difference between an 11/53,
11/73, 11/83 or 11/93 (or even 11/84 and 11/94). They all use the J-11.

But I think it is relevant to talk about the system, and not the CPU.
The 11/73 is different than an 11/93, even though they use the same CPU,
and you are emulating an 11/73.

TOY is definitely nice. Having some other clock solves the same problem,
making it less important, yes.

>>>>> And another interesting note. There are Digital Copyrights in all
>>>>> the source files. There are no AT&T Copyrights in any of the source
>>>>> files and only a few BSD Copyrights, Am I the only one who doubts
>>>>> that DEC rewrote the whole operating system in a clean room environment?
>>>>
>>>> People were a bit more sloppy back then.
>>>
>>> Not so sure it was sloppy. Wasn't this also a major point in the BSDI-AT&T
>>> lawsuit? Hadn't AT&T taken all of UCB's Copyright notices out of the code
>>> they used, like networking?
>>
>> It was, and yes. But was the removal done because they tried to hide the
>> origins, and pretend the code was done by them, or just because they
>> just wanted uniform file headers, and noone really gave much thought to
>> whether this was formally ok to do or not?
>> I think the latter, and I think that was just sloppy. But I might be wrong.
>
> Well, I thought the AT&T case was that someone said "AT&T software only
> has AT&T Copyrights" so the others were very deliberately removed. I
> suspect it was quite the same with DEC. And people think nobody understands
> copyright today!!

I doubt we'll even know the truth of the matter, but if they tried to
pull that stunt intentionally, then they obviously failed miserably.

But I think your comment can be read the same as mine about having
uniform file headers. And such a thing could have been executed without
anyone really thinking about the legality of the issue.

I guess it depends on whether the headers were changed prior to, or
after the lawsuit started appearing on the horizon.

Since DEC never even had a lawsuit about it, and still changed all the
headers, I think I'm leaning towards the "sloppy" argument there.

[Jerome H. Fine]

>Johnny Billquist wrote:

But, as far as CPU goes what other than speed did the 93 really offer?

TOY clock. Memory and serial ports on the same card as the CPU.

More specifically, comparing a PDP-11/93 with a PDP-11/83 and the
PDP-11/73, other than the small increase in speed, the big difference
was the number of boards required by the PDP-11/93 and the other
two systems.  For the PDP-11/83 with 2 * 2 MB boards of PMI memory,
four quad slots would be required to duplicate the CPU, memory and 8 DL
channels (on 2 * DLV11-J dual boards).  In a BA23 with just 8 quad
slots, that used half the available slots.  If more than 4 more quad slots
were required, the size of the final system could become a problem.
With a BA123 box with 12 quad slots, a PDP-11/83 system still had
8 quad slots left.

I don't have any dollar values for the original cost of a PDP-11/83 system,
but in the late 1990s, prices for the CPU. memory and DLV11-J boards
were substantially less than the single PDP-11/93 board.  As for the TOY
on the PDP-11/93 board, I doubt it would have been worth the extra cost.
As far as I know, the PDP-11/93 board is still much more expensive than
an equivalent set of boards for the PDP-11/73 or PDP-11/83 systems.

Finally, the speed difference between the three systems was relatively minor.
Very rough estimates using just DUP.SAV (to copy VM0: to itself) and
BINCOM.SAV (to compare VM0: with itself) suggest the the most important
factor was the PMI memory.  The faster speed of the PDP-11/93 CPU was
less of a factor.  The tests suggested that the PDP-11/83 was about 20% faster
than the PDP-11/73 (due to the PMI memory) and that the PDP-11/93
was about 33% faster than the PDP-11/73 (due to both the on-board
memory and the faster CPU) and just over 10% faster than the PDP-11/83.

I know that DEC provided a much higher evaluation, but those values
never seemed to be linked to what exactly was used as to hardware and
software.  If the PDP-11/93 used in the test had more memory and faster
disk drives, that would not have been a valid comparison in my opinion.

And as the risk of another comment from Johnny that I am repeating myself
with respect to the speed of Ersatz-11, here goes!

And if speed is a factor in the selection of the hardware for the PDP-11
system which is to be chosen and the only important criteria is the ability
to run the PDP-11 software (rather than the look, sound, smell and feel
of the hardware), then Ersatz-11 used with current Intel or AMD  CPUs
is somewhat faster than a PDP-11/93.  I did some recent tests on an
Intel system running a 64 bit Windows 7 and I estimate that the speed
was about 150 times a PDP-11/93.  The most recent 3rd party PDP-11
Qbus board from the Logical company also uses emulation to run the
software, although the final product is still a Qbus board that can also
be placed in a BA23 box (as far as I know).  I don't know of any exact
figures, but I would estimate that the speed is about 100 times a PDP-11/93.
But the cost starts at around $ 25,000 for the board. as compared with
about $ 6000 total if no other Qbus boards are required ($ 2000 for the
cost of a PC and a $ 4000 commercial license for Ersatz-11).  For
hobby users of the PDP-11, the only cost is the PC system if Ersatz-11
is used.

I doubt if anyone is interested in running any software on Ersatz-11, but
if there are any questions, please ask.

Jerome Fine

[Tom Lake]

"Jerome H. Fine" <ever...@nospam.com> wrote in message
news:50b2ada6$0$282$1472...@news.sunsite.dk...

>Johnny Billquist wrote: But, as far as CPU goes what other than speed did

>the 93 really offer?TOY clock. Memory and serial ports on the same card as
>the CPU.More specifically, comparing a PDP-11/93 with a PDP-11/83 and the

I run RSTS/E 10.0 on E11 every day (personal use only) and it’s great! There
are
three terminals hooked to serial ports (one Teletype ASR 33 and two Wyse
65s)
I’m teaching my kids BASIC and general programming techniques on the system
and we can all be using it at the same time.

Tom L

[glen herrmannsfeldt]

Tom Lake <tl...@twcny.rr.com> wrote:

(snip)

> I know that DEC provided a much higher evaluation, but those values
> never seemed to be linked to what exactly was used as to hardware and
> software. If the PDP-11/93 used in the test had more memory and faster
> disk drives, that would not have been a valid comparison in my opinion.

I would probably agree, but it reminds me that SPEC only benchmarks
systems, not CPUs.

So, yes, it doesn't provide a comparison for how much faster the CPU
is, but does for how much faster a system using that CPU, as it was
designed to be run, is.

-- glen

[Johnny Billquist]

Right. Running RT-11 you would not gain as much performance switching
from an 11/83 to an 11/93. One reason for the bigger gain claimed by DEC
is that the internal memory is totally off the bus, and much faster.
This is relevant when you have DMA and CPU executing at the same time,
as DMA can happen at the same time as the CPU accesses memory on the
11/93. On a single threaded thing like RT-11, who cares? You will
normally not have much I/O while doing computations.
It's a gain that you mostly realize when you have many things running at
the same time.
The same goes for intelligent disk controllers that can reorder I/O
operations. That too is pretty useless on RT-11. You will not, normally,
have an I/O queue to the controller holding three or more pending
requests at the same time.

Benchmarks were probably done with the same amount of memory. But they
were not done with RT-11. The only reason for any performance gain
between the 11/83 and 11/93 would be because all memory acts like cache
on the 11/93, while the 11/83 have cache on the CPU, but the hit rate is
probably around 90%.

Johnny

[Jerome H. Fine]

>Tom Lake wrote:

> I run RSTS/E 10.0 on E11 every day (personal use only) and it’s great!
> There are
> three terminals hooked to serial ports (one Teletype ASR 33 and two
> Wyse 65s)
> I’m teaching my kids BASIC and general programming techniques on the
> system
> and we can all be using it at the same time.

I run RT-11 95% of the time, but I do test some of the programs
under RSTS/E which are designed to run under RSTS/E. The
major problem I have is that I have not found how to transfer a
file directly into the RSTS/E 10.0 file structure for testing. I do
have a RSTS/E 7.0 system which supports a transfer between
RT-11 and RSTS/E after which I can MOUNT that device and
copy from RSTS/E 7.0 to RSTS/E 10.0, but I don't do it very
often. Does RSTS/E 10.0 have a transfer capability from RT-11
and if so, what command is used?

One aspect that is really helpful under E11 are the commands:

ASSIGN CON1: TT0:
ASSIGN CON2: TT1:
ASSIGN CON3: TT2:
etc.

which allows a user to directly access a DL line via the monitor
and keyboard on the PC. Obviously, it is possible to access only
one DL line at a time, but if you are using the DOS variant of E11,
<ALT/Fn> switches the monitor / keyboard to the desired CONn:
line. The Win32 variant of E11 also supports that capability, but
<ALT/Fn> may not work. Then just use the mouse to select the
CONn: that you want.

If you want a great game to teach your kids 2D geometry, try
SST.SAV under either RT-11 or RSTS/E.

One really great feature of E11 is that it supports a "terminal" size
other than just 24 lines by 80 columns. When you initiate:
E11 /CONSIZE:80x24
is the default, but E11 will support up to 255 columns by 60 lines
for the Win32 variant. For the DOS variant, the video card is
used to decide the different options. When you boot E11, use:
SHOW VIDEO
with the DOS variant to find out.

Jerome Fine

[Jerome H. Fine]

>Johnny Billquist wrote:

> Benchmarks were probably done with the same amount of memory. But they
> were not done with RT-11. The only reason for any performance gain
> between the 11/83 and 11/93 would be because all memory acts like
> cache on the 11/93, while the 11/83 have cache on the CPU, but the hit
> rate is probably around 90%.

I understand your point. Since the memory for the PDP-11/83
is not on the same board as the CPU, the path would be much
longer.

However, since the PDP-11/93 uses essentially the same CPU
and the PDP-11/83, I would assume that they both have the
same cache. Maybe that cache on the CPU could be called
the L1 cache for the PDP-11/93 and the on-board memory
designated the L2 cache which would be much closer to the
CPU. Because the memory is on the same board and the access
time is then so much less, an operating system like RSX-11
or RSTS/E could take better advantage.

Jerome Fine

[Johnny Billquist]

No. There is no cache on the 11/93. Instead all memory is high speed
memory, essentially the same thing as thinking that your CPU have 4M of
cache, and not the 8K of the 11/83.

Why do you assume they have the same cache just because they have the
same CPU? The cache is not on the CPU.

Johnny

[Jerome H. Fine]

>Johnny Billquist wrote:

> No. There is no cache on the 11/93. Instead all memory is high speed
> memory, essentially the same thing as thinking that your CPU have 4M
> of cache, and not the 8K of the 11/83.
>
> Why do you assume they have the same cache just because they have the
> same CPU? The cache is not on the CPU.

I appreciate your clarification. My assumption was that
the cache is on the J11 chip, but I appreciate the correction.

But just where is the cache? I have 3 different boards with
the J11 chip:
M8192 or KDJ11-AA
M8190 or KDJ11-BA
and a PDP-11/83 card which inaccessible in a system.

Since the cache is not on the J11 chip, I am still not able to
recognize it. Can you identify the location on the 3 boards
for me? I agree it is not an earth shattering problem if I
don't know, but I am curious.

Jerome Fine

[Johnny Billquist]

What exactly do you mean by "location"? You want the physical
coordinates on the KDJ11-B boards where the RAM chips sit?
I don't have any KDJ11-B boards nearby, so I can't provide information
at that level.
Search for small, fast RAM chips. There is a total of 8KB of cache
memory on the KDJ11-B board. I would guess something like 4114 chips,
but I might be very wrong on that one. There is a pretty detailed
description on how the cache works if you read the 11/84 manuals, but
they do not give the physical location of the memory chips on the board.

The same is true for the KDJ11-A.

The KDJ11-E however, do not have any cache. But like I said, DMA can
occur at the same time the CPU is accessing memory, since all memory is
used as cache, and the bus interface can be active at the same time as
the CPU is accessing memory, since the CPU access to memory does not at
all involve the bus.

[Johnny Billquist]

Oh, and the obvious speed gains from just having faster memory in an
11/93 compared to when you get cache misses in an 11/83 machine.

Cache misses are always costly.

[Johnny Billquist]

By the way. Found a KDJ11-A manual which states that chip E34 is the
cache control.

[Jerome H. Fine]

>Johnny Billquist wrote:

What exactly do you mean by "location"? You want the physical
coordinates on the KDJ11-B boards where the RAM chips sit?
I don't have any KDJ11-B boards nearby, so I can't provide information
at that level.
Search for small, fast RAM chips. There is a total of 8KB of cache
memory on the KDJ11-B board. I would guess something like 4114 chips,
but I might be very wrong on that one. There is a pretty detailed
description on how the cache works if you read the 11/84 manuals, but
they do not give the physical location of the memory chips on the board.

The same is true for the KDJ11-A.

The KDJ11-E however, do not have any cache. But like I said, DMA can
occur at the same time the CPU is accessing memory, since all memory is
used as cache, and the bus interface can be active at the same time as
the CPU is accessing memory, since the CPU access to memory does not at
all involve the bus.

Oh, and the obvious speed gains from just having faster memory in an 11/93 compared to when you get cache misses in an 11/83 machine.

Cache misses are always costly.

I remember that I wrote a small program at least 10 years ago
before I started using E11 for all my development.  The purpose
was to see how much faster cache was over memory.  I can't
find the program at the moment, but I suspect it was a simple
as placing a value into a block of memory using a tight loop,
then repeating the loop many times until the two loops took
many seconds.

The second test expanding the block of memory to double the
size so that it was larger than the cache, but then I cut the number
of outer loops in half.  When the first test was within cache and
the second test was too large for the cache, the second test took
much longer.  Otherwise, when both tests could operate within
cache, they took just about the same time.  And when both tests
exceeded the size of cache, they again took about the same time,
just much longer.

I was not looking for exact answers, just an approximation.
At this point, I can't remember the results at all.

Jerome Fine

[syts...@gmail.com]

On Thursday, November 22, 2012 3:18:49 PM UTC+1, Bill Gunshannon wrote:

> The networking has some bugs that need fixing, too.
>
>
>
> Is there any interest by others in playing with and updating Ultrix-11?
>
> Any particular things that you think need looking at? Is there even
>
> anyone else still running it?

Beginning of November, I've been playing with ultrix-11; I created an installed image, and moved that to my vhdl board - configured as 11/70 with rp06 and deuna. I was able to telnet to it, but sessions would get stuck after outputting as many bytes as the tcp window size. I'm not sure if that has anything to do with my deuna; it's microcode was still quite experimental at the time I ran the test. However, I could find no problem on my side of things, and the bug is quite peculiar - data is being transferred, but still some part of the OS seems to be unaware specifically of acknowledge counts. Anyway, I didn't have a clue where to look for a solution, so I changed the board to run 2.11bsd for a while - which it does brilliantly :-)

Maybe I can find some time around Xmas to continue with it.

[Bill Gunshannon]

In article <608cb0fb-af9d-4613...@googlegroups.com>,

I have seen that behavior as well, on incoming connections, both
telnet and ftp. But if I am on the console going out it works
fine. There is also a problem with ftp. If you are in binary
mode directory listing doesn't work. Amazing that no one saw
any of this when Ultrix-11 was still an active product.

I have Y2Ked the date command so that you can actually set the
clock.

I am in the process of setting up a system (I am using E11
at the moment but I do still have real hardware that I will
be able to test things on) with big disks to do development
on. One of the first things I want to do is get it set up
to make different kinds of install media and for that I am
hoping to add VTServer as a new device. It may be slow, but
it does work. I have installed BSD that way in the past.
And it is probably the most portable media most of us have
available to us.

I guess the big question will be what the new version will
be called, Will it be 3.2 or maybe just jump straight to
4.0 and go on from there. :-)

[Kuze Keisai]

On Thursday, 22 November 2012 14:18:49 UTC, Bill Gunshannon wrote:

> Is there any interest by others in playing with and updating Ultrix-11?
>
> Any particular things that you think need looking at? Is there even
>
> anyone else still running it?

Reading this,
I experience an insane and possibly unrealistic
desire to try running Ultrix 11 in XHomer DecPro emulator.

Beside Ultrix 4.5 RISC in GXEmul.

To get them to talk via network, even better.

[Johnny Billquist]

I don't think Ultrix-11 even supports the Pro...

[Bill Gunshannon]

In article <kur7ss$vbf$1...@iltempo.update.uu.se>,

Johnny Billquist <b...@softjar.se> writes:
> On 2013-08-18 19:26, Kuze Keisai wrote:
>> On Thursday, 22 November 2012 14:18:49 UTC, Bill Gunshannon wrote:
>>
>>> Is there any interest by others in playing with and updating Ultrix-11?
>>>
>>> Any particular things that you think need looking at? Is there even
>>>
>>> anyone else still running it?
>>
>> Reading this,
>> I experience an insane and possibly unrealistic
>> desire to try running Ultrix 11 in XHomer DecPro emulator.
>>
>> Beside Ultrix 4.5 RISC in GXEmul.
>>
>> To get them to talk via network, even better.
>
> I don't think Ultrix-11 even supports the Pro...
>

Nope.

[William Pechter]

In article <kur7ss$vbf$1...@Iltempo.Update.UU.SE>,

It doesn't -- however it will run Venix IIRC.
Getting Ultrix11 running would require major kernel hacking, I think
to get drivers to work.

Bill

--
--
Digital had it then. Don't you wish you could buy it now!
pechter-at-pechter.dyndns.org http://xkcd.com/705/

[Kuze Keisai]

Thank you for confirming.

I don't expect Ultrix 11 is DEC Pro ready,
and know it is probably impossible, or difficult.

But, I have XHomer and SIMH source,
so to me that's documentation
of the hardware differences.

Not promising anything,
but the way my mind works,
once I decide I want to do something,
even if it takes me five or ten years,
on and off loss of interest and effort,
I normally do it.

I don't need Ultrix on DECPro.
I have a nice fully working Ultrix 4.5 already,
as well as up to date Linux
on a low power Fujitsu laptop
that does for most things including 3DCG.

But, it is a romantic idea that arose :)

[Johnny Billquist]

On 2013-08-25 06:07, Kuze Keisai wrote:
> Thank you for confirming.
>
> I don't expect Ultrix 11 is DEC Pro ready,
> and know it is probably impossible, or difficult.

It is definitely not impossible. However, it's going to be damned hard.
Issues are both about needing a fair amount of new code, and possibly
some redesign, as the Pro differs from standard PDP-11 hardware design.
You'll also need to write a bunch of new code to deal with the graphics.

> But, I have XHomer and SIMH source,
> so to me that's documentation
> of the hardware differences.

Not sure how good the XHomer Pro emulation is, but I suspect it should
be enough.

> Not promising anything,
> but the way my mind works,
> once I decide I want to do something,
> even if it takes me five or ten years,
> on and off loss of interest and effort,
> I normally do it.

Well. Noone will stop you. :-)

> I don't need Ultrix on DECPro.
> I have a nice fully working Ultrix 4.5 already,
> as well as up to date Linux
> on a low power Fujitsu laptop
> that does for most things including 3DCG.
>
> But, it is a romantic idea that arose :)

Personally, I find Ultrix-11 to be a little too backward. It would be
way cooler to change XHomer to actually emulate a Pro-380, and then make
2.11BSD run on it.