7 Jan
2009
7 Jan
'09
9:38 a.m.
The vax unix nerds on the list might find this mildly interesting:
I hacked a copy of simh vax780 to look like a vax 11/730. I made a
modified version of the standalone netbsd "boot" program which does nfs
loads from a deuna (it will do copy's too, which is handy at times)
With that I finished debugging netbsd 2.0.3 on the 730. I've gotten it
to boot to a shell prompt on a simh 11/730 with 4mb of memory using an
nfs root. woo hoo!
So, next I'll make a disk image and put my unibus scsi card in my real
730 and debug that. With any luck it will just work.
Once it's going on real hardware I'll get it going on the latest netbsd
(4.0?) and get a patch back to the netbsd/vax maintainer.
[for those who care, I use a linux program which simulates the tu58 to
boot my 730. I made a boot tape image with putr which contains the
netbsd boot program (as well as the other files needed to load
microcode, etc). I plan to spend some time and make a linux program
which will put together valid tu58 tape images, just to make my life
easier, since putr requires DOS.]
I have to give the netbsd folks a lot of credit. Their posix build
system is very handy. But note that releases before 2.0.3 will *NOT*
build on a posix machine - they require native netbsd to build cleanly.
I've always had a soft spot for the 730; maybe because the 780 was
just too damn big! :-)
-brad
7 Jan
7 Jan
10:28 a.m.
On 7 Jan 2009 at 12:38, Brad Parker wrote:
I've always had a soft spot for the 730; maybe
because the 780 was
just too damn big! :-)
I can recall doing a price-performance analysis of VAXen around 1983
or so. My conclusion was that the 750 was the most bang for the buck
and that the 730 was pretty underpowered for the money. Just curious
if this was really the case. At any rate, we were pretty happy with
the 750.
Cheers,
Chuck
11:18 a.m.
-----Original Message-----
From: cctalk-bounces at classiccmp.org [mailto:cctalk-
bounces at classiccmp.org] On Behalf Of Brad Parker
Sent: Wednesday, January 07, 2009 9:38 AM
To: General Discussion: On-Topic and Off-Topic Posts
Subject: got netbsd 2.0.3 to boot on 11/730 in sim
The vax unix nerds on the list might find this mildly interesting:
I hacked a copy of simh vax780 to look like a vax 11/730. I made a
modified version of the standalone netbsd "boot" program which does nfs
loads from a deuna (it will do copy's too, which is handy at times)
With that I finished debugging netbsd 2.0.3 on the 730. I've gotten it
to boot to a shell prompt on a simh 11/730 with 4mb of memory using an
nfs root. woo hoo!
Very cool - congrats!
[snip]
I've always had a soft spot for the 730; maybe because the 780 was
just too damn big! :-)
[snip]
Too big? C'mon, it's only 1200 lbs. (550 kg) or so....
As an interesting coincidence, I just got a VAX-11/780-5 to boot into VMS (yesterday!).
After doing a careful restoration of the potentially broken bits (power supplies and
such), I got the machine running diagnostics. N.B.: you need to be sure to have the
*right* diagnostics for your machine and its components. Just because the diskette says
it's a diagnostic for an 11/785 doesn't mean it will match your hardware. The
available literature on the versioning leaves something to be desired; I think DEC assumed
they'd give you a set of floppies matched to your machine and you'd be happy
evermore. (If you're undertaking this and want to know more about what I learned,
please drop me an email.)
Getting VMS going was a challenge because of peripherals. The machine has a RH780, but we
are using our RP06s on our PDP-10s; our RP07s all need restoration; and we don't have
enough power in the room where the VAX lives to run one. The machine also has a CI780, so
I suggested using a hierarchical storage controller. We briefly considered a HSC, but
decided against investing in RA drives. :-) So we picked up an HSJ50 from PSDS in
Woodinville, Washington - really great people for old DEC stuff. They built it out for us
with a full complement of RZ29s in Storageworks shelves - which was a mixed blessing....
When we decided to go the HSJ route, I figured we'd have a controller, perhaps a
single shelf and then be able to plug in a CDROM and/or magtape drive into the unused
ports. The system PSDS built for us is FULLY built out, with six shelves and RAID arrays
done properly - vertically, across the shelves. So I had no free ports. I tried a couple
of simple ideas, not wanting to completely disassemble the system, but I was facing the
idea of restructuring the RAID arrays to free up a shelf, unplugging that shelf from the
HSJ and plugging in the CDROM. Yuck.
What I ended up doing was easier and worked out well. I used a MicroVAX (3100) setup with
a single Storageworks shelf and a CDROM drive. I booted the MicroVAX into standalone
backup and restored the install save set onto an RZ29; I also copied the entire CD onto
another RZ. I reduced a couple of RAID storagesets on the HSJ, installed the RZs from the
MicroVAX and was able to boot from then!
There was one interesting bit of learning curve on this. I was unable to get standalone
backup to boot on the VAX-11/780-5, despite having a set of disks for that purpose: the
VAX-11 never 'saw' the stuff on the HSJ. I verified that CI was working correctly
(while SABU was running I disabled the CI ports at the HSJ and SABU complained), but still
couldn't access devices. So I tried simply booting, using CIBOO.CMD on the console
disk. One challenge was figuring out the device number: R2 contains the CI node(s) and R3
contains the unit number at that node (or in the case of our dual-redundant setup, nodes).
It wasn't clear what that number would be: SCSI ID, shelf ID, some conflation of the
two?
I finally figured out something that wasn't clear from the HSJ documentation: the unit
IDs are NOT arbitrary labels. Keep in mind that disks are represented as:
disk(s) ---> storageset (RAID, mirror, etc.) ---> unit
When I received the storage system, things were set up with disk IDs like DISK100,
representing shelf 1, SCSI ID 0, LUN 0; RAID IDs like R0, R1, etc.; and unit IDs of D0, D1
and so forth. After some futzing, I figured out: the unit ID numbers (digits) ARE the IDs
to be deposited in the VAX-11's R3 register. So to boot D6 (one of the single-disk
units I'd added), R3 <-- 6. Happy dance!
So I'm going to rebuild the RAID arrays with full membership and install onto one.
I'm going to have more storage than a VAX-11 knows what to do with. :-) And once
that's all done, I have a VAX-11/785 that's still ill - yes, we have TWO. Is that
cool or what? Cheers -- Ian
4:25 p.m.
New subject: Crossing compiling HP3000 MPE on a VAX was Re: got netbsd 2.0.3 to boot on 11/730 in sim
Hi Brad!
Topic drift, but your post reminded me of one of the most interesting uses for a VAX I
know of.
Back when HP was developing PA-RISC (mid-1980's) to replace the 16-bit stack
architecture 3000 we used a VAX-750 running BSD to host a PA-RISC simulator. I was working
on the MPE boot-strap code at the time and used the simulator to write in PA-RISC
assembler (no HLL compiler yet) and run tests. We had one 750 for the entire MPE OS Lab,
although only a few of us were using it. Not surprising that the turn-around for compiles
and debugging reminded me of mainframe batch turn-arounds from school, at the end of the
semester of course. Barely usable.
We soon got proto-type CPUs implemented in TTL that plugged into the back-plane of a very
heavily hacked Series 44 (desk form-factor 3000). That way we could have hardware'ish
speed and use real peripherals. Huge improvement before we got real prototype hardware to
do OS bring-up on.
Of course this was all very hush-hush at the time since HP was locked in a death struggle
with DEC in the mini-computer market. Would have been bad press for it to get out that HP
used DEC computers to design its new 32-bit computer architecture.
Cheers,
Lee Courtney
--- On Wed, 1/7/09, Brad Parker <brad at heeltoe.com> wrote:
From: Brad Parker <brad at heeltoe.com>
Subject: got netbsd 2.0.3 to boot on 11/730 in sim
To: "General Discussion: On-Topic and Off-Topic Posts" <cctalk at
classiccmp.org>
Date: Wednesday, January 7, 2009, 9:38 AM
The vax unix nerds on the list might find this mildly
interesting:
I hacked a copy of simh vax780 to look like a vax 11/730.
I made a
modified version of the standalone netbsd "boot"
program which does nfs
loads from a deuna (it will do copy's too, which is
handy at times)
With that I finished debugging netbsd 2.0.3 on the 730.
I've gotten it
to boot to a shell prompt on a simh 11/730 with 4mb of
memory using an
nfs root. woo hoo!
So, next I'll make a disk image and put my unibus scsi
card in my real
730 and debug that. With any luck it will just work.
Once it's going on real hardware I'll get it going
on the latest netbsd
(4.0?) and get a patch back to the netbsd/vax maintainer.
[for those who care, I use a linux program which simulates
the tu58 to
boot my 730. I made a boot tape image with putr which
contains the
netbsd boot program (as well as the other files needed to
load
microcode, etc). I plan to spend some time and make a
linux program
which will put together valid tu58 tape images, just to
make my life
easier, since putr requires DOS.]
I have to give the netbsd folks a lot of credit. Their
posix build
system is very handy. But note that releases before 2.0.3
will *NOT*
build on a posix machine - they require native netbsd to
build cleanly.
I've always had a soft spot for the 730; maybe because
the 780 was
just too damn big! :-)
-brad
8 Jan
8 Jan
10 a.m.
New subject: HP's use of DEC equipment RE: Crossing compiling HP3000 MPE on a VAX
From: Lee Courtney
Sent: Wednesday, January 07, 2009 4:25 PM
Topic drift, but your post reminded me of one of the
most interesting uses
for a VAX I know of.
Back when HP was developing PA-RISC (mid-1980's)
to replace the 16-bit stack
architecture 3000 we used a VAX-750 running BSD to host a PA-RISC simulator.
[big snip ]
Of course this was all very hush-hush at the time
since HP was locked in a
death struggle with DEC in the mini-computer market. Would have been bad
press for it to get out that HP used DEC computers to design its new 32-bit
computer architecture.
Hi, Lee,
[I didn't know you had worked for HP. You may have known one of my student
friends from Stanford who worked on HP-UX for PA-RISC after he graduated, but
that's a topic for another time.]
This was not the first time, or first DEC architecture, that HP used in product
development. Several of the engineers at XKL were former DEC-20 users while
employed at HP, which used the SUDS CAD package (as did cisco Systems, where
many of them went from HP). Some of my old friends were the system managers
for those machines.
Rich Alderson
Server Engineer, PDPplanet Project
Vulcan, Inc.
505 5th Avenue S, Suite 900
Seattle, WA 98104
mailto:RichA at vulcan.com
(206) 342-2239
(206) 465-2916 cell
11:33 a.m.
New subject: HP's use of DEC equipment RE: Crossing compiling HP3000 MPE on a VAX
--- On Thu, 1/8/09, Rich Alderson <RichA at vulcan.com> wrote:
From: Rich Alderson <RichA at vulcan.com>
Subject: HP's use of DEC equipment RE: Crossing compiling HP3000 MPE on a VAX
To: "'General Discussion: On-Topic and Off-Topic Posts'" <cctalk at
classiccmp.org>
Date: Thursday, January 8, 2009, 10:00 AM
From: Lee Courtney
Sent: Wednesday, January 07, 2009 4:25 PM
Send me a name(s) off-line. I wasn't heavily engaged with the HP-UX Lab, but my wife
was (and still is).
Topic drift, but your post reminded me of one of
the
most interesting uses
for a VAX I know of.
Back when HP was developing PA-RISC
(mid-1980's)
to replace the 16-bit stack
architecture 3000 we used a VAX-750 running BSD
to
host a PA-RISC simulator.
[big snip ]
Hi, Lee,
[I didn't know you had worked for HP. You may have
known one of my student
friends from Stanford who worked on HP-UX for PA-RISC
after he graduated, but
that's a topic for another time.] This was not the first time, or first DEC
architecture,
that HP used in product
development. Several of the engineers at XKL were former
DEC-20 users while
employed at HP, which used the SUDS CAD package (as did
cisco Systems, where
many of them went from HP). Some of my old friends were
the system managers
for those machines.
Very interesting - if you know more I'd be interested in hearing (off-line).
When I joined HP in the late 70s I had used TOPS-20 for several years, but had never
touched an HP box.
For a while the hardware Lab also had several 3270 type terminals used for CAD/CAM,
connected to one of the IBM mainframes in Palo Alto. By the time the PA-RISC Lab moved
from Building 42 to 44 in Cupertino, and the project transitioned from HP-Labs to the
product division I believe all that work had moved to HP workstations.
We also had tricked out IBM AS400 living in the MPE Lab for a while, but that's a
different story...
Along the same lines I also heard a rumor that in the 1960s SDS used an IBM mainframe for
its internal accounting, even as they were competing with the 360 in the business DP
marketplace.
Interesting how different vendors infiltrated others. Often so things could get done
faster and more efficiently.
Cheers,
Lee Courtney
27 Jan
27 Jan
5:49 p.m.
New subject: HP's use of DEC equipment RE: Crossing compiling HP3000 MPE on a VAX
Re:
We also had tricked out IBM AS400 living in the MPE
Lab for a while,
Circa 1979/1980, the MPE lab (building 47U) had an IBM 8100 (IIRC) in one of
the small conference rooms (southeast corner). I remember seeing an IBM CE
come in to repair the memory after the HP engineers zapped it for the second or
third time :)
In this case, I think it was being used for reverse-engineering / competitive
analysis, not for it's intended use.
Stan
--
Stan Sieler
sieler at allegro.com
www.allegro.com/sieler/wanted/index.html
5812
days inactive
5833
days old
6 comments
6 participants
participants (6)
-
brad＠heeltoe.com -
cclist＠sydex.com -
charlesleecourtney＠yahoo.com -
IanK＠vulcan.com -
RichA＠vulcan.com -
sieler＠allegro.com