PORTS/EPORTS @ <300 bps?
SDF Poster
DoN. Nichols
> As anyone successfully run a PORTS or EPORTS board at less than 300 bps?
PORTS or EPORTS board? Was that the COMBO board with no RAM
installed, just the two serial ports?
I don't really remember a PORTS or EPORTS board for the
7300/3b1/unix-pc.
Perhaps refresh my memory on this?
And of course, the question arises of "Why?". :-) I've used 110
baud modems and Teletypes, and my feeling is *never again*. :-)
If I had a system up, I could look as the choices of parameters
to see whether anything slower than 300 baud was supported. They may
have been stolen to use for higher baud rates such as 19200 (which was
really stressing the system. :-)
Enjoy,
DoN.
--
Email: <dnic...@d-and-d.com> | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---
SDF Poster
> PORTS or EPORTS board? Was that the COMBO board with no RAM
>installed, just the two serial ports?
The PORTS and EPORTS cards are feature boards for the 3B2. I realize
that 3b1/7300 users aren't 3B users and infact, they're only slightly
related. However, comp.sys.att has been dead and comp.sys.3b1 has not.
I have indeed confirmed that just defining a 110 line in gettydefs is
all that is needed to communicate with equipment at 110 baud.
> And of course, the question arises of "Why?". :-) I've used 110
>baud modems and Teletypes, and my feeling is *never again*. :-)
The ASR 33 only does 110. But I initially tried it out with a Model 43
(10 cps / 30 cps) first since that would be the most reliable way of
testing. BTW, this is for an exhibit at a Western Electric Museum. Next
step is to get the ASCII to BAUDOT/current converter working.
But don't despair, there is a 3B1 on exhibit and available for use at
the museum. Once I remove the battery and clean it up a bit I'll see
if it too can communicate at 110 baud over its eia port.
DoN. Nichols
> DoN. Nichols <dnic...@d-and-d.com> wrote:
>> PORTS or EPORTS board? Was that the COMBO board with no RAM
>>installed, just the two serial ports?
>
> The PORTS and EPORTS cards are feature boards for the 3B2.
Aha! I thought that they sounded familiar (and AT&T) - but I
couldn't find in reference to it in the 3b1 documentation. :-)
> I realize
> that 3b1/7300 users aren't 3B users and infact, they're only slightly
> related.
Both sold by AT&T, and both ran a SysV unix -- and the only
difference between the two names is the '1' or '2' after the "3B", so I
guess that will do.
> However, comp.sys.att has been dead and comp.sys.3b1 has not.
O.K. Welcome aboard. I suspect that more people were able to
afford the 3B1 on a hobby basis than the 3B2.
> I have indeed confirmed that just defining a 110 line in gettydefs is
> all that is needed to communicate with equipment at 110 baud.
Good enough.
>> And of course, the question arises of "Why?". :-) I've used 110
>>baud modems and Teletypes, and my feeling is *never again*. :-)
>
> The ASR 33 only does 110. But I initially tried it out with a Model 43
> (10 cps / 30 cps) first since that would be the most reliable way of
> testing. BTW, this is for an exhibit at a Western Electric Museum. Next
> step is to get the ASCII to BAUDOT/current converter working.
O.K. For that, it certainly makes sense.
Hmm ... there was a version of the monitor ROM for my first
computer -- the Altair 680b -- which would talk to a Baudot terminal.
Of course, I went the hard way, and wrote my own driver to print on a
Frieden Flexowriter -- with a totally different characterset, and used
an ASCII keyboard for input -- at least until a failure in the
Flexowriter fried most of the computer. I got it working again, but
with a ADM-3a terminal (kit built) instead.
> But don't despair, there is a 3B1 on exhibit and available for use at
> the museum.
Where is the museum? New Jersey, perhaps?
> Once I remove the battery and clean it up a bit I'll see
> if it too can communicate at 110 baud over its eia port.
The best thing with the battery is to get a clip for a coin cell
for the 2032 for a pretty good fit. This reduces the soldering-iron
trauma for subsequent battery changes. This is what I did with the 3B1.
And I believe that the 110 baud works just as well there. It
may be the 50 baud which was stolen to get 19200 instead.
SDF Poster
> Both sold by AT&T, and both ran a SysV unix -- and the only
>difference between the two names is the '1' or '2' after the "3B", so I
>guess that will do.
Well, to me they were very different. The 3b1 having a Motorola 68000
and the 3b2 having a WE32000. SVR3.5x has the Version 7 UUCP package,
I believe as well as a few other differences that make it seem that the
68k port started out life much earlier than the 3b1 or 7300 hit the
market. Of course, there is 'ua' too. The 3b2/300 came out with
SVR5r2 and the BNU package used the later version of UUCP. 3b2s did
support ST506, but also ESDI and SCSI .. so, to me, and even though I
don't know the full story on both, they're from different corners of
the warehouse at the very least ;-)
> O.K. Welcome aboard. I suspect that more people were able to
>afford the 3B1 on a hobby basis than the 3B2.
I've actually been here since about 1992, thank you! :) Yes, the 3b1/7300
was much more economical when it was released compared to the 16k - 24k a
base 3b2 would have put you back. However, it was dog slow .. though even
at the time you could swap in a faster (and larger) drive.
> Where is the museum? New Jersey, perhaps?
We're in Seattle WA, near Georgetown. http://museumofcommunications.org
> The best thing with the battery is to get a clip for a coin cell
>for the 2032 for a pretty good fit. This reduces the soldering-iron
>trauma for subsequent battery changes. This is what I did with the 3B1.
I did that on a 7300 in 1994. It had already drained by then. That
particular system now had bad video memory (another problem that comes
unnannounced). I do have memory chips, sockets .. now I just need time.
> And I believe that the 110 baud works just as well there. It
>may be the 50 baud which was stolen to get 19200 instead.
Interesting, that is good to know. The 3B1 definitely will show up
on the Starlan to do UUCP. One of the supporters of the museum who is
now deceased donated his 'road kit' which included several field test
systems and parts which he used servicing 3B2s, 3B1/7300s and 6300+ I
personally did System 75/85 Audix service for a while including the WGS
(x86) systems running Audix as well.
DoN. Nichols
> DoN. Nichols <dnic...@d-and-d.com> wrote:
>
>> Both sold by AT&T, and both ran a SysV unix -- and the only
>>difference between the two names is the '1' or '2' after the "3B", so I
>>guess that will do.
>
> Well, to me they were very different. The 3b1 having a Motorola 68000
> and the 3b2 having a WE32000.
Also -- the 3b1/7300/unix-pc was actually designed and built by
Convergent Technologies.
> SVR3.5x has the Version 7 UUCP package,
O.K. The HDB UUCP was made available for the 3B1 later, and I
installed it.
> I believe as well as a few other differences that make it seem that the
> 68k port started out life much earlier than the 3b1 or 7300 hit the
> market.
Or it may be that Convergent was limited as to which licenses
they had while doing the early developing. But it wound up (by version
3.51) rolling together features from SysVr2, SysVr3, and BSD. A pity
that it did not pick up the BSD FFS (Fast File System) which allowed for
file names in excess of 14 characters.
> Of course, there is 'ua' too. The 3b2/300 came out with
> SVR5r2 and the BNU package used the later version of UUCP. 3b2s did
> support ST506, but also ESDI and SCSI
With plug in cards, not on the system board as the ST506 on the
3B1 was implemented.
Note that there *was* a SCSI card for the 3B1 -- but it was
never released. One of the regulars here had one (perhaps the only
prototype). The company developed it just in time of the commercial
failure of the 3B1, so it was kept in-house.
> .. so, to me, and even though I
> don't know the full story on both, they're from different corners of
> the warehouse at the very least ;-)
I don't know who actually constructed the 3B2s -- but it was
Convergent who made the 3B1s.
>> O.K. Welcome aboard. I suspect that more people were able to
>>afford the 3B1 on a hobby basis than the 3B2.
>
> I've actually been here since about 1992, thank you! :) Yes, the 3b1/7300
> was much more economical when it was released compared to the 16k - 24k a
> base 3b2 would have put you back. However, it was dog slow .. though even
> at the time you could swap in a faster (and larger) drive.
You have not *seen* dog slow. I had (and still have, though
long retired) a Cosmos CMS-16/UNX. A v7 unix port to the 68000 (not
68010, which was needed to support virtual memory). The CPU was an 8MHz
one, compared to the 10 MHz 68010 in the 3b1. Just for the fun of it,
here are some dhrystone benchmarks on the various systems:
======================================================================
* MACHINE MICROPROCESSOR OPERATING COMPILER DHRYSTONES/SEC.
* TYPE SYSTEM NO REG REGS
* -------------------------- ------------ ----------- ---------------
* Cosmos 68000-8Mhz UniSoft cc 305 322 Report with this distribution
* Cosmos 68000-8Mhz UniSoft cc 365 394 - My machine (Plessy ram cards)
* ATT 3B2/300 WE32000-?Mhz UNIX 5.0.2 cc 735 806
* ATT PC7300 68010-10Mhz UNIX 5.2 cc 1041 1111
* ATT PC7300 68010-10Mhz UNIX 5.2(3.51) cc 1066 1137
* Sun2/120 68010-10Mhz Standalone cc 1219 1315
* ATT 3B2/400 WE32100-?Mhz UNIX 5.2 cc 1315 1315
And for the other end of things, my current fastest system :-)
* SunFire 280R 2x900MHz 64 Solaris 10-U3 gcc 4.2.0 1199040 1210653 (slow -- why?)
* SunFire 280R 2x900MHz 64 Solaris 10-U3 cc 5.9 2145922 2145922
* SunFire 280R 2x900MHz 32 Solaris 10-U3 gcc 4.2.0 2173913 2183406
* SunFire 280R 2x900MHz 32 Solaris 10-U3 cc 5.9 2314814 2304147
======================================================================
Note the difference in speed made in the Cosmos by different
memory cards. The second entry was my own tests on my system, the other
came with the dhrystone benchmark source.
The 3B2/300 was noticeably slower than the 3B1 with the version
3.51 OS. The Sun2/120 was even faster, with the same 10 MHz 68010.
The Cosmos was a Unisoft port of both the v7 OS and the
compiler, and that compiler treated the 68000 as being the same as a
PDP-11 (except for the link/unlink instructions for building/destroying
stack frames). Looking at the assembly code produced by the compiler
showed it to be not using things like the single-instruction array
indexing built into the 68K family, while the same source showed that
the 3B1's compiler (at the 3.51 OS version, which supposedly had a
faster compiler) did.
As another comparison, I got a Tektronix 6130 (NS 16032 CPU
IIRC) with no root password, but with a guest account which let me read
the passwd file, so I built a combined version of the /usr/dict/words
file from all three systems, and wrote a program which tried each word
as the password, using the salt in the password entry, and with the
entry hard-coded into the program for speed. The 3B1 was the fastest of
the three (I set it running on all three), and eventually found that the
password was "gnome", so I could finally do serious things with the
machine. It was actually this difference which prompted me to compare
the assembly source produced by the compilers. :-)
>> Where is the museum? New Jersey, perhaps?
>
> We're in Seattle WA, near Georgetown. http://museumofcommunications.org
O.K. A long distance from where I had guessed. :-)
>> The best thing with the battery is to get a clip for a coin cell
>>for the 2032 for a pretty good fit. This reduces the soldering-iron
>>trauma for subsequent battery changes. This is what I did with the 3B1.
>
> I did that on a 7300 in 1994. It had already drained by then. That
> particular system now had bad video memory (another problem that comes
> unnannounced). I do have memory chips, sockets .. now I just need time.
Hmm ... I had not encountered that problem -- but what I would
suggest prior to replacing the video memory is to check on the power
supply connector and the underside of the power supply where the
connector pins solder on. The 5V was split into multiple pins, and each
pin served a different subsystem, so an oxidized pin could introduce
different problems. One would corrupt disks during writes (as well as
slow reads), another would crash the system, and I think that a third
fed the video portion of the system.
>> And I believe that the 110 baud works just as well there. It
>>may be the 50 baud which was stolen to get 19200 instead.
>
> Interesting, that is good to know. The 3B1 definitely will show up
> on the Starlan to do UUCP.
O.K. I never used Starlan, but did fire up the ethernet to tall
to my mix of other systems.
> One of the supporters of the museum who is
> now deceased donated his 'road kit' which included several field test
> systems and parts which he used servicing 3B2s, 3B1/7300s and 6300+
That would be interesting to see.
You know tha the 6300 had even worse problems than the Y2K which
hit the 3B1. It had an internal clock which only spanned eight years,
after which you needed to install a replacement driver to add the proper
value to the clock to come up for the current eight year span. A pity
that they didn't simply write it to read the start year from a file. :-)
> I
> personally did System 75/85 Audix service for a while including the WGS
> (x86) systems running Audix as well.
Hmm ... given the voice power cards available for the 3B1, did
anyone ever build an Audix system using them?
SDF Poster
> Also -- the 3b1/7300/unix-pc was actually designed and built by
>Convergent Technologies.
I think they also built a 386 very cool looking block tower on casters
too. Maybe it was 68k. If so, then yes, that would have been even
cooler.
> O.K. The HDB UUCP was made available for the 3B1 later, and I
>installed it.
Yes, as did I. You probably know that Taylor was even ported to the
3b2 and 3b1.
>that it did not pick up the BSD FFS (Fast File System) which allowed for
>file names in excess of 14 characters.
Yes. I had an internal release of SVR4 for the 3B2 but it was
unfortunately lost years ago. Thankfully a 3B2 FTP server showed up for
a brief stint (I think run by Rob Kas) which had disk images. However,
I wouldn't dare run it for production. Even SVR4 on the 386 was a bit
unstable (Sorry if I offend any SVR4 fans .. if you actually exist).
As far as I'm concerned (and maybe 3B2ers) SVR3.2.3 2.0 was the 'last
version of UNIX'.
> The 3B2/300 was noticeably slower than the 3B1 with the version
>3.51 OS. The Sun2/120 was even faster, with the same 10 MHz 68010.
Do you have an original 300? You have a 400, so you know what the MMU,
MAU and CPU look like. If the 300 chips look very different, then you
have a very early version. Sometimes you can tell you've got the early
version if the bottom half of the case is black rather than dark brown.
> O.K. I never used Starlan, but did fire up the ethernet to tall
>to my mix of other systems.
I used STARLAN 1 w/ RFS extensively. These days I'm using NetBSD NIS+NFS
and while its of course faster and can handle tens of thousands of users,
I remember how rock solid RFS was. But those are religious fighting
words... don't respond.
> That would be interesting to see.
I'll get photos of the road kit for the 6300+ .. its at one of the COs
in the basement. I could pick it up sometime next week.
> Hmm ... given the voice power cards available for the 3B1, did
>anyone ever build an Audix system using them?
We probably have those but if not, its definitely something that could
be donated and put into use by the Museum.
DoN. Nichols
> DoN. Nichols <dnic...@d-and-d.com> wrote:
>> Also -- the 3b1/7300/unix-pc was actually designed and built by
>>Convergent Technologies.
>
> I think they also built a 386 very cool looking block tower on casters
> too. Maybe it was 68k. If so, then yes, that would have been even
> cooler.
Hmm ... they had a server and a bunch of clients powered by
Intel chips (I never saw the server so I don't know what it looked like,
a co-worker got some of the clients, and we could never make them work
without knowing the protocols the server used.
But they also had the miniframe -- a 68020 based bigger brother
to the 3B1s. I've only read about that too -- never seen it.
>> O.K. The HDB UUCP was made available for the 3B1 later, and I
>>installed it.
>
> Yes, as did I. You probably know that Taylor was even ported to the
> 3b2 and 3b1.
Taylor? I'm afraid that I did not know about that. Was that
yet another implementation of uucp?
>>that it did not pick up the BSD FFS (Fast File System) which allowed for
>>file names in excess of 14 characters.
>
> Yes. I had an internal release of SVR4 for the 3B2 but it was
> unfortunately lost years ago. Thankfully a 3B2 FTP server showed up for
> a brief stint (I think run by Rob Kas) which had disk images. However,
> I wouldn't dare run it for production. Even SVR4 on the 386 was a bit
> unstable (Sorry if I offend any SVR4 fans .. if you actually exist).
Well ... I'm pretty happy with Sun's Solaris 10, which is based
on SysVr4, but lots of Sun additions. The early Solaris versions, (up
through perhaps 2.5) were rather slow, but 2.6 up through 10 (which
should have been Solaris 2.10 if they had kept the old naming scheme
going) have been getting faster and more powerful. I'm particularly
impressed with Sun's zfs (Zetabyte File System) which is built into
Solaris 10. It is a very nice RAID implementation, with lots of extra
data checksumming. I'm running two zfs pools at the moment, one with
five 36 GB Fibre Channel (FC) drives (and two hot spares), and the other
with five 18 GB SCA SCSI drives, and one hot spare, which is soon going
to migrate to another array of FC drives. I discovered that I could
migrate an array, one drive at a time, using the "zpool replace"
command, while the filesystems on that array were in constant use by my
wife and me.
> As far as I'm concerned (and maybe 3B2ers) SVR3.2.3 2.0 was the 'last
> version of UNIX'.
:-)
I'm also using OpenBSD 4.2 on Sun Ultra machines, FWIW. I got
into Suns with the 2/120, which was a BSD based OS, and it remained BSD
based until SunOs 4.1.4. The first SysV sun was called Solaris 2.0, and
the combination of SunOs 4.1.x and X11 with the Openwindows window
manager was called Solaris 1.y during the transition to SysVr4 based
Solaris.
These days, Solaris 10 will only run on the UltraSPARC systems,
and on i386 systems. This means that I have some older systems running
Solaris 2.6, and just recently retired a SunOs 4.1.4 system.
>> The 3B2/300 was noticeably slower than the 3B1 with the version
>>3.51 OS. The Sun2/120 was even faster, with the same 10 MHz 68010.
>
> Do you have an original 300? You have a 400, so you know what the MMU,
> MAU and CPU look like.
No. I never had a 3B2. The only 3B2 hardware I have is the box
which held two disk drives and a Floppy Tape drive (a terrible tape
implementation, that) which I adapted to hold the same sort of drives
for a heavily modified 3B1. One of the benefits is that the color scheme
was a pretty good match. I was also able to steal the 5V from the power
connector for the original internal hard drive and use it to control the
switch in the 3B2 disk chassis' power supply.
Had to have a machinist friend make a drive mount plate for the
second drive, as it was not present. (I could do it at home, now, but
I did not have the machines then.)
> If the 300 chips look very different, then you
> have a very early version. Sometimes you can tell you've got the early
> version if the bottom half of the case is black rather than dark brown.
Nope -- I don't have them at all.
>> O.K. I never used Starlan, but did fire up the ethernet to tall
>>to my mix of other systems.
>
> I used STARLAN 1 w/ RFS extensively. These days I'm using NetBSD NIS+NFS
> and while its of course faster and can handle tens of thousands of users,
> I remember how rock solid RFS was. But those are religious fighting
> words... don't respond.
:-)
>> That would be interesting to see.
>
> I'll get photos of the road kit for the 6300+ .. its at one of the COs
> in the basement. I could pick it up sometime next week.
>
>> Hmm ... given the voice power cards available for the 3B1, did
>>anyone ever build an Audix system using them?
>
> We probably have those but if not, its definitely something that could
> be donated and put into use by the Museum.
The reason that I asked was that I picked up at a hamfest an
expansion chassis for the 3B1 which was mostly populated with Voice
Power boards. The system to which it had been attached was gone, but at
least they had pulled the interface board and kept it with the expansion
chassis.