Hi,

Yesterday I rebuilt Tops-10 7.04 from the Operating System and compiler load tapes using simh's KL-10 emulation. The total sysgen took about 1 hour 20 minutes on my Linux machine running on my old laptop Intel i7-4900MQ.

I wanted to pose this question to any of the original computer operators who did a PDP-10 Tops-10 7.03/7.04 SYSGEN from tapes on a real PDP-10 KL-10 as to how long this SYSGEN of the OS + loading of compilers from tape would have taken on the real KL-10 hardware.

Thanks,
Bill

William Sternbach <wgs777@gmail.com> writes:

> Hi,

> Yesterday I rebuilt Tops-10 7.04 from the Operating System and compiler load
> tapes using simh's KL-10 emulation. The total sysgen took about 1 hour 20
> minutes on my Linux machine running on my old laptop Intel i7-4900MQ.

> I wanted to pose this question to any of the original computer operators who
> did a PDP-10 Tops-10 7.03/7.04 SYSGEN from tapes on a real PDP-10 KL-10 as to
> how long this SYSGEN of the OS + loading of compilers from tape would have
> taken on the real KL-10 hardware.

Hi, Bill,

I last did a full sysgen of Tops-10 v7.04 on real hardware in 2005, after we
got the XKL KL-10 up and running at Paul Allen's facility (later the site of
Living Computers: Museum + Labs).

As I recall, it took two or three hours to build the monitor, but some of that
time was spent running the MONGEN program to set up the final monitor build.
It would have taken much longer if the monitor had to be assembled from source.

The result was not bootable, by the way; the issue was that the people who put
the thing together made the assumption that there was a working monitor already
in place which MONGEN would run under, and local commands defined for one of
the tables which MONGEN built. Since I was running the minimal monitor from
the bootable tape, and had no local commands to define, the resulting monitor
ended up going into uninitialized parts of memory when booted.

The monitor which people used on that KL-10 when we created the online museum
(PDPplanet.com) in 2005 and later after the renaming to LivingComputerMuseum.org
in 2008 (eventually shortened to LivingComputers.org) was the fully functional
monitor from the KLAD installation, modified in two places using FILDDT: The
name field typed out at login, and the value of the monitor variable INDPPN
(changed from 0 to -1) which made every project,programmer number independent,
instead of every programmer number with the same project having full access to
all files under the shared PPN.

--
Rich Alderson news@alderson.users.panix.com
Audendum est, et veritas investiganda; quam etiamsi non assequamur,
omnino tamen proprius, quam nunc sumus, ad eam perveniemus.
--Galen

On Saturday, September 17, 2022 at 9:04:12 PM UTC-4, William Sternbach wrote:
> Hi,
>
> Yesterday I rebuilt Tops-10 7.04 from the Operating System and compiler load tapes using simh's KL-10 emulation. The total sysgen took about 1 hour 20 minutes on my Linux machine running on my old laptop Intel i7-4900MQ.
>
> I wanted to pose this question to any of the original computer operators who did a PDP-10 Tops-10 7.03/7.04 SYSGEN from tapes on a real PDP-10 KL-10 as to how long this SYSGEN of the OS + loading of compilers from tape would have taken on the real KL-10 hardware.
>
> Thanks,
> Bill

Hi Rich,

Thank you for your post and the interesting info about a sysgen on real hardware taking 2 to 3 hours.
So emulating the KL-10 on simh is faster than the real hardware was for the heavy I/O based sysgen process.
I'm using Rich Cornwell's Tops 10 7.04 load script on his simh emulator. He also ran into the same problem that you
ran into with the monitor. I believe his script uses a tops10 7.03 monitor at first to get the 7.04 sysgen started.

My only exposure to the PDP-10 was in 1983 during my first computer science class in freshman year in College
in the programming language Lisp, which was a killer class that I had to do constant all nighter's on just to survive.
I continued to use the PDP-10 in my sophmore year (1984) until they eventually and sadly got rid of the PDP-10
and replaced it with a Vax 11/785. I remember walking through the small building which housed the computer room.
It had a thick glass wall, so everyone could see the operators at work and you could see the real hardware.

Using Tops-10 made such a lasting impression on me that nearly 40 years later I am still admiring what a great OS it was.
I use my Tops-10 7.04 emulated build on simh like my primary computer daily. I use the SOS editor on Tops-10 to
edit files making daily "to do" lists. I've been reading a lot of the Dec Tops-10 original compiler manuals from the 1970's and 1980's
(particularily the Cobol 74 and Fortran manual) and writing some programs on it daily.
I do it mainly for it's historical value, to be using the original operating system, editor, compilers etc, that most universities were
using in the 1970's and 1980's.

I believe the DEC OS and compilers were so superior to the IBM equivalent IBM 370 / 4341 computers running either MVS or VM/CMS
at that time, that DEC really could have given away some free samples to the largest companies in America and demoed how superior
their OS and computers were, and could have taken on IBM. But sadly, Dec's management seemed to be satisfied with just the
University / Research set of customers, and then sadly cancelled the Jupiter project. So many times throughout history, something so good
that it has a loyal following generations later got cancelled by short sighted management.

I would like to visit the living computer museum on a vacation sometime. Would like to see and study the real hardware
of a PDP-10 KL-10. I'd like to show it to my children (3 of them are in college now), to show them what computers were like
when I was their age.

Thanks again for your response.

Bill

On Friday, September 23, 2022 at 2:42:53 AM UTC-7, William Sternbach wrote:

(snip)

> My only exposure to the PDP-10 was in 1983 during my first computer science class in freshman year in College
> in the programming language Lisp, which was a killer class that I had to do constant all nighter's on just to survive.
> I continued to use the PDP-10 in my sophmore year (1984) until they eventually and sadly got rid of the PDP-10
> and replaced it with a Vax 11/785. I remember walking through the small building which housed the computer room.
> It had a thick glass wall, so everyone could see the operators at work and you could see the real hardware.
We also had a PDP-10 in college, though a little earlier. It was a KA-10 running TOPS-10,
and got replaced by two VAX 11/780 in about 1979. No glass wall, or even window, so I
never actually saw the KA-10 or VAX.

Then a few years later, I was working in a place with a TOPS-20 system, that got
replaced with three VAX 11/750 in about 1983 or 1984. They sold the TOPS-20
system, so it went away very fast.

LCM closed at the beginning of the Covid shelter in place, and so far
has not reopened.

gah4 wrote:
> LCM closed at the beginning of the Covid shelter in place, and so far
> has not reopened.

Clarification: closed to regular visitors. Still open for remote
logins: ssh menu@tty.livingcomputers.org

One of which was the last MIT PDP-10 running ITS. Still running ITS,
thanks to LCM. (Seems unresponsive right now, though.)

On Friday, September 23, 2022 at 11:04:11 AM UTC-4, lars...@nocrew.org wrote:
> gah4 wrote:
> > LCM closed at the beginning of the Covid shelter in place, and so far
> > has not reopened.
> Clarification: closed to regular visitors. Still open for remote
> logins: ssh me...@tty.livingcomputers.org
>
> One of which was the last MIT PDP-10 running ITS. Still running ITS,
> thanks to LCM. (Seems unresponsive right now, though.

The PDP-10 seems to have made such a lasting impression on so many people.
Each person has their own favorite OS, whether its ITS, tops-10, or tops-20.
A sign of a really great machine that even 40 to 50 years later it still has a loyal following.

Regarding the PDP-10, are there any DEC PDP-10 KL-10B's (Not Toads) still online and running?
If so, would it be possible to request a login to them?
I have some benchmark programs that I wrote in Cobol and Fortran which I've ran on my KL-10 tops-10 v7.04 image with both emulators: simh and klh10.
Would be very interesting to run them on the real hardware to compare cpu time used on real hardware versus the emulators.

William Sternbach <wgs777@gmail.com> writes:

> On Friday, September 23, 2022 at 11:04:11 AM UTC-4, lars...@nocrew.org wrote:
>> gah4 wrote:

>>> LCM closed at the beginning of the Covid shelter in place, and so far
>>> has not reopened.

>> Clarification: closed to regular visitors. Still open for remote
>> logins: ssh me...@tty.livingcomputers.org

>> One of which was the last MIT PDP-10 running ITS. Still running ITS,
>> thanks to LCM. (Seems unresponsive right now, though.

> The PDP-10 seems to have made such a lasting impression on so many people.
> Each person has their own favorite OS, whether its ITS, tops-10, or tops-20.
> A sign of a really great machine that even 40 to 50 years later it still has
> a loyal following.

> Regarding the PDP-10, are there any DEC PDP-10 KL-10B's (Not Toads) still
> online and running? If so, would it be possible to request a login to them?
> I have some benchmark programs that I wrote in Cobol and Fortran which I've
> ran on my KL-10 tops-10 v7.04 image with both emulators: simh and klh10.
> Would be very interesting to run them on the real hardware to compare cpu
> time used on real hardware versus the emulators.

The obfuscated login above is "menu@tty...". Look for the DEC-10 entry. The
system *may* be running; I don't know.

New accounts are no longer available on the KL-10, but you can certainly have a
look. Check out Kermit for file transfers--there was no TCP/IP networking for
Tops-10, and DECnet is not installed on the system.

--
Rich Alderson news@alderson.users.panix.com
Audendum est, et veritas investiganda; quam etiamsi non assequamur,
omnino tamen proprius, quam nunc sumus, ad eam perveniemus.
--Galen

On Monday, September 26, 2022 at 6:13:37 PM UTC-4, Rich Alderson wrote:
> William Sternbach <wgs...@gmail.com> writes:
>
> > On Friday, September 23, 2022 at 11:04:11 AM UTC-4, lars...@nocrew.org wrote:
> >> gah4 wrote:
>
> >>> LCM closed at the beginning of the Covid shelter in place, and so far
> >>> has not reopened.
>
> >> Clarification: closed to regular visitors. Still open for remote
> >> logins: ssh me...@tty.livingcomputers.org
>
> >> One of which was the last MIT PDP-10 running ITS. Still running ITS,
> >> thanks to LCM. (Seems unresponsive right now, though.
>
> > The PDP-10 seems to have made such a lasting impression on so many people.
> > Each person has their own favorite OS, whether its ITS, tops-10, or tops-20.
> > A sign of a really great machine that even 40 to 50 years later it still has
> > a loyal following.
>
> > Regarding the PDP-10, are there any DEC PDP-10 KL-10B's (Not Toads) still
> > online and running? If so, would it be possible to request a login to them?
> > I have some benchmark programs that I wrote in Cobol and Fortran which I've
> > ran on my KL-10 tops-10 v7.04 image with both emulators: simh and klh10.
> > Would be very interesting to run them on the real hardware to compare cpu
> > time used on real hardware versus the emulators.
> The obfuscated login above is "menu@tty...". Look for the DEC-10 entry. The
> system *may* be running; I don't know.
>
> New accounts are no longer available on the KL-10, but you can certainly have a
> look. Check out Kermit for file transfers--there was no TCP/IP networking for
> Tops-10, and DECnet is not installed on the system.
> --
> Rich Alderson ne...@alderson.users.panix.com
> Audendum est, et veritas investiganda; quam etiamsi non assequamur,
> omnino tamen proprius, quam nunc sumus, ad eam perveniemus.
> --Galen

Thanks Rich - I'll take a look.

On Monday, September 26, 2022 at 3:13:37 PM UTC-7, Rich Alderson wrote:
> William Sternbach <wgs...@gmail.com> writes:

(snip)

> > Regarding the PDP-10, are there any DEC PDP-10 KL-10B's (Not Toads) still
> > online and running? If so, would it be possible to request a login to them?
> > I have some benchmark programs that I wrote in Cobol and Fortran which I've
> > ran on my KL-10 tops-10 v7.04 image with both emulators: simh and klh10.
> > Would be very interesting to run them on the real hardware to compare cpu
> > time used on real hardware versus the emulators.

> The obfuscated login above is "menu@tty...". Look for the DEC-10 entry. The
> system *may* be running; I don't know.
Unless I am reading it wrong, the KA and KL are emulated.

The KS is running ITS.

On Saturday, September 17, 2022 at 6:04:12 PM UTC-7, William Sternbach wrote:

> Yesterday I rebuilt Tops-10 7.04 from the Operating System and compiler load tapes using simh's KL-10 emulation. The total sysgen took about 1 hour 20 minutes on my Linux machine running on my old laptop Intel i7-4900MQ.

> I wanted to pose this question to any of the original computer operators who did a PDP-10 Tops-10 7.03/7.04 SYSGEN from tapes on a real PDP-10 KL-10 as to how long this SYSGEN of the OS + loading of compilers from tape would have taken on the real KL-10 hardware.
There are some who try to make emulators run close to the time of the original system.

That has been especially true for video game playing systems, that are unplayable at today's speed.

In the case of sysgen, you need both processor clock and I/O speeds to run slower.

One interesting problem in getting old OS to run on emulated hardware,
is hardware that runs too fast. In some cases, programs made assumptions
on how long something would take, especially I/O. When the I/O operation
is done immediately, some assumption fails.

I don't know of any of those for TOPS-10, though.

You could almost have the emulator compute the time that it would take,
even though it is running fast. That is complicated by overlap between
CPU and I/O, though I suspect you get close by just I/O time.

On Tuesday, October 4, 2022 at 12:57:07 PM UTC-4, gah4 wrote:
> On Saturday, September 17, 2022 at 6:04:12 PM UTC-7, William Sternbach wrote:
>
> > Yesterday I rebuilt Tops-10 7.04 from the Operating System and compiler load tapes using simh's KL-10 emulation. The total sysgen took about 1 hour 20 minutes on my Linux machine running on my old laptop Intel i7-4900MQ.
>
> > I wanted to pose this question to any of the original computer operators who did a PDP-10 Tops-10 7.03/7.04 SYSGEN from tapes on a real PDP-10 KL-10 as to how long this SYSGEN of the OS + loading of compilers from tape would have taken on the real KL-10 hardware.
> There are some who try to make emulators run close to the time of the original system.
>
> That has been especially true for video game playing systems, that are unplayable at today's speed.
>
> In the case of sysgen, you need both processor clock and I/O speeds to run slower.
>
> One interesting problem in getting old OS to run on emulated hardware,
> is hardware that runs too fast. In some cases, programs made assumptions
> on how long something would take, especially I/O. When the I/O operation
> is done immediately, some assumption fails.
>
> I don't know of any of those for TOPS-10, though.
>
> You could almost have the emulator compute the time that it would take,
> even though it is running fast. That is complicated by overlap between
> CPU and I/O, though I suspect you get close by just I/O time.

Very true.
I did a near perfect KL-10B sysgen with simh from the great resources at Rich Cornwell's website.

I then backed up the pdp-10 disk files after the sysgen (so I won't need to do sysgens again),
and I also applied the klh10 patch to allow klh10 to be able to sence when cpu is idle.
The allows klh10 to use only .08% of cpu time on a modern linux system.

I have 1 master Tops-10 build now which can run smoothly on either emulator..
klh10 uses less CPU time and runs synthetic benchmarks faster than simh, but simh has a lot more needed functionality
like kermit file transfers, tape functionality, and multi terminal functionality since Simh Implements a multi-serial adapter DZ10, and its channels are routed to telnet.

On Monday, November 14, 2022 at 5:32:52 AM UTC-8, William Sternbach wrote:

> klh10 uses less CPU time and runs synthetic benchmarks faster than simh, but simh has a lot more needed functionality
> like kermit file transfers, tape functionality, and multi terminal functionality since Simh Implements a multi-serial adapter DZ10, and its channels are routed to telnet.

I'm not sure I understand this. I run ITS under KLH10, and I can do kermit, I have tape functionality, and I can run many simultaneous telnet/supdup-based sessions.

Eric Swenson <eric@swenson.org> writes:

> On Monday, November 14, 2022 at 5:32:52 AM UTC-8, William Sternbach wrote:

>> klh10 uses less CPU time and runs synthetic benchmarks faster than simh, but
>> simh has a lot more needed functionality like kermit file transfers, tape
>> functionality, and multi terminal functionality since Simh Implements a
>> multi-serial adapter DZ10, and its channels are routed to telnet.

> I'm not sure I understand this. I run ITS under KLH10, and I can do kermit, I
> have tape functionality, and I can run many simultaneous telnet/supdup-based
> sessions.

Standard Tops-10 does not support TCP/IP networking.

--
Rich Alderson news@alderson.users.panix.com
Audendum est, et veritas investiganda; quam etiamsi non assequamur,
omnino tamen proprius, quam nunc sumus, ad eam perveniemus.
--Galen