While watching Biography about Bill Gates on CNBC last Night

Daddio

unread,

Dec 10, 2009, 7:05:38 PM12/10/09

to

One of the pictures they had of Bill Gates showed a DEC20 or a DEC10
Mainframe computer from Digital Equipment corporation. That I used
to work on for many many years.

My my those were the days

Mikey

--
--------------------------------- --- -- -
Posted with NewsLeecher v4.0 Beta 10
Web @ http://www.newsleecher.com/?usenet
------------------- ----- ---- -- -

Andrew McLaren

unread,

Dec 12, 2009, 2:45:17 AM12/12/09

to

Daddio wrote:
> One of the pictures they had of Bill Gates showed a DEC20 or a DEC10
> Mainframe computer from Digital Equipment corporation. That I used
> to work on for many many years.

The DEC PDP-10 and DECSYSTEM-20 were admirable machines; but they don't
qualify as "mainframes". They were (are?) minicomputers.

Arguably, the only "mainframe" produced by DEC was the VAX 9000.

This is especially true in the late 70s, early 80s when "real"
mainframes were clearly characterised by "massive" IO channels (some of
those suckers could move 17MBps).

Cheers
Andrew

--
amclar at optusnet dot com dot au

jmfbahciv

unread,

Dec 12, 2009, 9:39:18 AM12/12/09

to

Andrew McLaren wrote:
> Daddio wrote:
>> One of the pictures they had of Bill Gates showed a DEC20 or a DEC10
>> Mainframe computer from Digital Equipment corporation. That I used to
>> work on for many many years.
>
> The DEC PDP-10 and DECSYSTEM-20 were admirable machines; but they don't
> qualify as "mainframes". They were (are?) minicomputers.

WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG

>
> Arguably, the only "mainframe" produced by DEC was the VAX 9000.

WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG

>
> This is especially true in the late 70s, early 80s when "real"
> mainframes were clearly characterised by "massive" IO channels (some of
> those suckers could move 17MBps).
>

We've discussed this before and you've still not learned.

/BAH

Bill Leary

unread,

Dec 12, 2009, 11:48:26 AM12/12/09

to

OK, Barb.

Please illuminate for someone who was there and involved in buying these
things, but didn't catch your earlier discussion.

When we were shopping machines, the DEC marketroid and the engineers we had
in called them "minicomputers" and occasionally "mini-mainframes." Why is a
DEC PDP-10 not a minicomputer?

- Bill

"jmfbahciv" <jmfbahciv@aol> wrote in message
news:hg097...@news3.newsguy.com...

Andrew McLaren

unread,

Dec 12, 2009, 5:36:13 PM12/12/09

to

jmfbahciv wrote:
>
> We've discussed this before and you've still not learned.

Au contraire! I learnt lots of things. But DEC computers still don't
qualify for the term 'mainframe' :-)

The perplexing bit is that some Digital folks still clamour for the
supposed prestige of the "mainframe" label; as though something they
deserved, had been unfairly withheld from them. In some circles,
"mainframe" is a term of abuse! Who'd want it?

FWIW I love DEC computers, still have a VAXStation 4000 stacked away in
the shed, and a copy of Kenah and Goldberg in my bookshelf. Oh, and I
like mainframes, too :-)

Cheers
Andrew

jmfbahciv

unread,

Dec 13, 2009, 9:15:15 AM12/13/09

to

Bill Leary wrote:
> OK, Barb.
>
> Please illuminate for someone who was there and involved in buying these
> things, but didn't catch your earlier discussion.
>
> When we were shopping machines, the DEC marketroid and the engineers we
> had in called them "minicomputers" and occasionally "mini-mainframes."

That would have been for the VAX system, not the -10.

> Why is a DEC PDP-10 not a minicomputer?
>

Because it wasn't. It was a mainframe and was sold as one. Nobody
ever called it a mini. We called the VAXes as Midi, a tad more
powerful (and bigger) than the minis. VAXes didn't even run
as a mainframe until VMS started to transform into a genuine
timesharing system.

/BAH

jmfbahciv

unread,

Dec 13, 2009, 9:19:35 AM12/13/09

to

Andrew McLaren wrote:
> jmfbahciv wrote:
>>
>> We've discussed this before and you've still not learned.
>
> Au contraire! I learnt lots of things. But DEC computers still don't
> qualify for the term 'mainframe' :-)

Then you don't know anything about the biz back then. PDP-10s were
not, and never were, minis. Even the KA-10 with 48K core running
4S72 was not called a mini. The PDP-8Is that hung off the front
of it were called minis.

>
> The perplexing bit is that some Digital folks still clamour for the
> supposed prestige of the "mainframe" label;

It had nothing to do with prestige. It had to do with how the
damned system was used.

>as though something they
> deserved, had been unfairly withheld from them.

Sigh! You are talking about the fucking VMS mess not the PDP-10
biz.

>In some circles,
> "mainframe" is a term of abuse! Who'd want it?

thus showing your fucking bias. I want the functionality,
versatility and performance of a mainframe.

>
> FWIW I love DEC computers, still have a VAXStation 4000 stacked away in
> the shed, and a copy of Kenah and Goldberg in my bookshelf. Oh, and I
> like mainframes, too :-)

Obviously, you wouldn't know a mainframe if it bit you in the keester.

/BAH

Jorgen Grahn

unread,

Dec 13, 2009, 9:11:26 AM12/13/09

to

On Sat, 2009-12-12, Andrew McLaren wrote:
...

> The perplexing bit is that some Digital folks still clamour for the
> supposed prestige of the "mainframe" label; as though something they
> deserved, had been unfairly withheld from them. In some circles,
> "mainframe" is a term of abuse! Who'd want it?

s/some/most/, surely?

I haven't read much about that part of history, but it seems to me
Digital is usually portraited as the guys who liberated companies from
IBM serfdom by building open and cheap computers mini computers. Or
bringing digital computing to organizations who were too small to be
interesting to Big Blue.

Or maybe that's just my Unix bias showing.

/Jorgen

--
// Jorgen Grahn <grahn@ Oo o. . .
\X/ snipabacken.se> O o .

Anne & Lynn Wheeler

unread,

Dec 13, 2009, 9:49:05 AM12/13/09

to

jmfbahciv <jmfbahciv@aol> writes:
> Because it wasn't. It was a mainframe and was sold as one. Nobody
> ever called it a mini. We called the VAXes as Midi, a tad more
> powerful (and bigger) than the minis. VAXes didn't even run
> as a mainframe until VMS started to transform into a genuine
> timesharing system.

vaxs & 43xx sold into similar/same midrange market. price and
price/performance seem to break some threshhold and there was huge
increase in uptake in the volumes. large customers even had orders of
43xx in numbers of multiple hundreds at a time. part of it was some
number of datacenters were bursting at the seams & 43xx (& vaxs) could
be installed in areas with minimum preparation (i.e. some locations
seeing disappearing conference rooms as they were being converted to
computer rooms).

the 43xx followons (4361s & 4381s) had expected to see similar volume
uptakes ... but by that time, much of that mid-range market was starting
to move to workstations and large PCs (servers) ... which can also be
seen in the vax numbers ... old post giving vax volumes sliced & diced
by year, us/non-us, model, etc
http://www.garlic.com/~lynn/2002f.html#0 Computers in Science Fiction

future system was going to completely replace 360/370 ... and
that distraction led to drying up 370 (hardware & software) product
pipeline, in the wake of failure of future system
http://www.garlic.com/~lynn/submain.html#futuresys

there was mad rush to get products back into the 370 product
pipelines. part of that was appetite to pickup a lot of 370 stuff I had
been doing all during the future system period. another part was big
push to define 370-xa and get it out (which was going to take
7-8yrs). the pok group managed to convince corporate that vm370 needed
to be killed and the (burlington mall) vm370 development group shutdown,
and all the people moved to POK in order to make mvs/xa schedule. In
any case, quite a few people didn't move.

One of the jokes is the head of POK was major contributor to VMS
(because of the number of people from the burlington mall group showing
up at DEC). With the increasing uptake in midrange (at the time
138/148), endicott manage to assume the vm370 product responsibility
... but had to reconsitute a development group from scratch.

old email in this post
http://www.garlic.com/~lynn/2001m.html#15 departmental servers
referring to 43xx multi-hundred order
http://www.garlic.com/~lynn/2001m.html#email790404b

other old email mentioning 43xx
http://www.garlic.com/~lynn/lhwemail.html#4341

recent posts about work on cluster scaleup was facilitated by
being able to port over RDBMS that had vaxcluster support
http://www.garlic.com/~lynn/2009b.html#40 "Larrabee" GPU design question
http://www.garlic.com/~lynn/2009b.html#43 "Larrabee" GPU design question
http://www.garlic.com/~lynn/2009h.html#26 Natural keys vs Aritficial Keys
http://www.garlic.com/~lynn/2009k.html#36 Ingres claims massive database performance boost
http://www.garlic.com/~lynn/2009k.html#67 Disksize history question
http://www.garlic.com/~lynn/2009m.html#57 Ikea type font change
http://www.garlic.com/~lynn/2009p.html#35 DB2 announces technology that trumps Oracle RAC and Exadata
http://www.garlic.com/~lynn/2009p.html#38 big iron mainframe vs. x86 servers
http://www.garlic.com/~lynn/2009p.html#42 big iron mainframe vs. x86 servers
http://www.garlic.com/~lynn/2009p.html#43 From The Annals of Release No Software Before Its Time
http://www.garlic.com/~lynn/2009p.html#46 From The Annals of Release No Software Before Its Time
http://www.garlic.com/~lynn/2009p.html#49 big iron mainframe vs. x86 servers
http://www.garlic.com/~lynn/2009p.html#54 big iron mainframe vs. x86 servers
http://www.garlic.com/~lynn/2009q.html#26 Check out Computer glitch to cause flight delays across U.S.

--
40+yrs virtualization experience (since Jan68), online at home since Mar1970

Bill Leary

unread,

Dec 13, 2009, 12:19:07 PM12/13/09

to

"jmfbahciv" <jmfbahciv@aol> wrote in message

news:hg2s6...@news4.newsguy.com...

> Bill Leary wrote:
>> OK, Barb.
>>
>> Please illuminate for someone who was there and involved in buying these
>> things, but didn't catch your earlier discussion.
>>
>> When we were shopping machines, the DEC marketroid and the engineers we
>> had in called them "minicomputers" and occasionally "mini-mainframes."
>
> That would have been for the VAX system, not the -10.

OK. That got me to thinking. The company I was with later bought the VAX
without having anyone come into the office. We did everything by phone.
The software we wanted to run required a VAX so there wasn't any debate
about what we were going to buy, just how it was configured. So it can't
have been VAX that I'm remembering.

But...

>> Why is a DEC PDP-10 not a minicomputer?
>>
> Because it wasn't. It was a mainframe and was sold as one. Nobody
> ever called it a mini.

Generally, I'd argue "Because it wasn't." is no answer. But you seem so
sure, I did a bit of research on it before arguing further. Especially
examining the machine system architecture, I find I have to agree.

Also, again looking over how these were built and sold, it seems most
unlikely the company I was with at the time I remember dealing with the DEC
marketing and engineering departments would have been trying to buy one of
these. I suspect I've made a memory slip and they were trying to sell us
PDP-11's.

I think the problem with accepting this for most people is that our thoughts
of PDP are dominated by the other models, and especially the 11. I wonder
if that might not have been a marketing handicap at the time as well?

- Bill

Peter Flass

unread,

Dec 13, 2009, 1:58:51 PM12/13/09

to

Anne & Lynn Wheeler wrote:
>
> vaxs & 43xx sold into similar/same midrange market.

I don't believe I ever heard the 4341 call a mini, it was considered a
small mainframe. The later 9370 was a mini, and was entirely
rack-mountable.

In my opinion, the 4341 was one of the best machines ever made. The
price/performance was great and the reliability was excellent (AFAIK).
After that, the direction of IBM was steadily downhill.

We had an outfit called BRS (Bibliographic Retrieval Services) in
Schenectady that ran a couple of 4341's for text search and retrieval,
sort of the Google of their day, running two 4341's. They used STAIRS
to search, I believe, medical information. They were a spinoff of the
stuff we were doing at SUNY Research Foundation searching for grant
information.

Ryan McCoskrie

unread,

Dec 14, 2009, 6:51:47 AM12/14/09

to

Jorgen Grahn wrote:

I think it is your bias there 'cause even I'm familiar with that old story
and I've never seen more than a smallish rack of servers[1]. If a
technological legend is that well known then it has to be old opinions being
passed on.

[1] I'm only eighteen here people.
--
Quote of the login:
[We] use bad software and bad machines for the wrong things. -- R.W. Hamming

Esra Sdrawkcab

unread,

Dec 14, 2009, 8:15:13 AM12/14/09

to

On Sun, 13 Dec 2009 18:58:51 -0000, Peter Flass <Peter...@yahoo.com>
wrote:

STAIRS! thankyou, that's the term I've been looking for for the past 3
weeks - it indexed *all* words, IIRC.

--
Nuns! Reverse!

jmfbahciv

unread,

Dec 14, 2009, 8:49:37 AM12/14/09

to

Ryan McCoskrie wrote:
> Jorgen Grahn wrote:
>
>> On Sat, 2009-12-12, Andrew McLaren wrote:
>> ...
>>
>>> The perplexing bit is that some Digital folks still clamour for the
>>> supposed prestige of the "mainframe" label; as though something they
>>> deserved, had been unfairly withheld from them. In some circles,
>>> "mainframe" is a term of abuse! Who'd want it?
>> s/some/most/, surely?
>>
>> I haven't read much about that part of history, but it seems to me
>> Digital is usually portraited as the guys who liberated companies from
>> IBM serfdom by building open and cheap computers mini computers. Or
>> bringing digital computing to organizations who were too small to be
>> interesting to Big Blue.
>>
>> Or maybe that's just my Unix bias showing.
>
> I think it is your bias there

Nope. What he said is true. The comments I heard in the Mill was
that IBM wanted DEC to be in business because DEC kept IBM from
being considered a monopoly.

> 'cause even I'm familiar with that old story
> and I've never seen more than a smallish rack of servers[1]. If a
> technological legend is that well known then it has to be old opinions being
> passed on.
>
> [1] I'm only eighteen here people.

Welcome :-).

11's weren't used as servers back then; there would be PDP-8s. After
the DC72 product, the DC76 and DC78 products were sold; those were
PDP-11s. There was also the DAS78 but that was a specially made
system. All of the above eventually became the product we called
ANF-10 and started to be distributed on the TOPS-10 monitor tape
as a bundled product. When this happened the command
SET HOSTESS was implemented. CDO called the (I think it was a)
DC76 TWINKIE so he could type the command

SET HOSTESS TWINKIE.

Now, I've given you oodles of googles; go read about them. :-)

/BAH

jmfbahciv

unread,

Dec 14, 2009, 8:51:14 AM12/14/09

to

Anne & Lynn Wheeler wrote:

> jmfbahciv <jmfbahciv@aol> writes:
>> Because it wasn't. It was a mainframe and was sold as one. Nobody
>> ever called it a mini. We called the VAXes as Midi, a tad more
>> powerful (and bigger) than the minis. VAXes didn't even run
>> as a mainframe until VMS started to transform into a genuine
>> timesharing system.
>
> vaxs & 43xx sold into similar/same midrange market. price and
> price/performance seem to break some threshhold and there was huge

<snip>

A mainframe is a stand-alone machine unless the OS makes it perform
like a mainframe.

/BAH

Walter Bushell

unread,

Dec 14, 2009, 9:33:46 AM12/14/09

to

In article <op.u4xi3nxjhswpfo@dell3100>,
"Esra Sdrawkcab" <ad...@127.0.0.1> wrote:

Even articles and propositions^W prepositions?

--
A computer without Microsoft is like a chocolate cake without mustard.

Walter Bushell

unread,

Dec 14, 2009, 9:34:37 AM12/14/09

to

In article <hg5f5...@news3.newsguy.com>, jmfbahciv <jmfbahciv@aol>
wrote:

????

Anne & Lynn Wheeler

unread,

Dec 14, 2009, 9:41:55 AM12/14/09

to

Peter Flass <Peter...@Yahoo.com> writes:
> I don't believe I ever heard the 4341 call a mini, it was considered a
> small mainframe. The later 9370 was a mini, and was entirely
> rack-mountable.

re:
http://www.garlic.com/~lynn/2009r.html#33 While watching Biography about Bill Gates on CNBC last Night

it wasn't about what it was called ... it was what market it was being
sold into. some bought it as a datacenter machine ... but midrange was a
lot of machines outside the traditional datacenter environment,
sometimes being purchased hundreds at a time for distributed operation.

ibm user group SHARE ... saw vax/vms having some competitive advantage
(in some of that market) because of the perceived lower skill & resource
effort to install and operate.

the datacenter issue ... referred to here
http://www.garlic.com/~lynn/2001m.html#15 departmental servers

was that cluster of 4341s had lower cost, higher aggregate mip rate,
more storage and more channel that 3033. this is separate from
4341 having significantly better price/performance than 3031

as mentioned in the above post ... there was talk by head of POK that
claimed that there were 11,000-plus VAX sales that should have been
4341s (4341 having better price/performance ... but as share implied,
pure price/performance isn't always the only criteria).

there was also some internal (dept. of dirty tricks) politics. giving
some credability to the perceived threat of 4341 to (high-end/pok) 303x
machines ... a plant under POK control ... that manufactured a critical
component for 4341, at one point, was directed to cut the
allocation/supply to endicott (for 4341 manufacturing) in half.

I did some benchmarking for the endicott 4341 engineers against 158 &
3031. The issue was that I was (also) getting to play disk engineer in
bldgs. 14&15, as well as supplying them with highly enhanced operating
system so they could do engineering & test in operating system
environment (they had been running stand-alone, some tests with running
MVS in that environment had a 15min MTBF). In any way, the disk labs
would get early processor models for testing/validating that disk worked
with the machines (in addition to using the machines to testing new
disks & controllers). The net was, that I had better access to 4341 (as
well as 145, 158, 3031, & 3033) for testing than the performance testing
group in Endicott. misc. past posts mentioning getting to play disk
engineer
http://www.garlic.com/~lynn/subtopic.html#disk

jmfbahciv

unread,

Dec 14, 2009, 10:01:45 AM12/14/09

to

Do you know what a stand-alone machine is?

/BAH

Walter Bushell

unread,

Dec 14, 2009, 10:05:08 AM12/14/09

to

In article <hg5j9...@news1.newsguy.com>, jmfbahciv <jmfbahciv@aol>
wrote:

Oh, I thought you meant, "A mainframe is a stand-alone machine unless
the OS makes it perform like a micro". (Or perhaps a mini.)

Anne & Lynn Wheeler

unread,

Dec 14, 2009, 10:09:48 AM12/14/09

to

re:
http://www.garlic.com/~lynn/2009r.html#33 While watching Biography about Bill Gates on CNBC last Night

http://www.garlic.com/~lynn/2009r.html#37 While watching Biography about Bill Gates on CNBC last Night

some of the relationship with endicott (and the science center) ...
some past science center posts
http://www.garlic.com/~lynn/subtopic.html#545

had started before 370 virtual memory was announced. a joint project
between endicott & science center (including vnet connection that was
used to send source code changes back&forth) involving modifying cp67
(running on 360/67) to simulate 370 (virtual memory) virtual machine
(370 had some new/changed instructions that didn't exist on 360/67 and
the hardware virtual memory tables were different between 360/67 and
370).

a big part of the simulate 370 virtual memory virtual machine was doing
all the operations that took information from the (virtual hardware)
tables in the 370 virtual machine and translating it into 360/67
formated "shadow" tables (i.e. 370 information translated into 360
format ... and virtual machine page numbers converted to the "real" page
numbers).

this was running in regular use at cambridge a year before "real" 145
hardware supporting virtual memory was even operational (and well before
announce). part of the issue was that science center had a lot of users
from various educational institutions in the boston/cambrige area
(including students) ... and there was security concerns about 370
virtual memory information leaking out. as a result ... the modified
cp67 (supporting 370 virtual machine option) wasn't normally run on the
real hardware (system used by non-employees) ... but in a 360/67 virtual
machine. Then there was a cp67 that modified to run on "real" 370
(instead of 360/67) ... which ran in a 370 virtual machine. Sort of
normal operation:

1) real 360/67
2) "standard" cp67 running on real 360/67
3) "H" modified cp67 running in 360/67 virtual machine
4) "I" modified cp67 running in 370 virtual machine
5) cms running in virtual machine

this was in normal operation a year before real 370 (virtual memory)
hardware was available. In fact the "cp67-i" system (designed to run on
real 370 ... was used in initial test the 370/145 engineering machine
(with virtual memory hardware). The "joke" was that initial IPL/boot
didn't work ... after some investigation ... it turned out that the
hardware engineers had reversed the op-codes for some of the new 370
instructions. The cp67-i system was then patched to use the (incorrect)
opcodes and then things booted succesfully.

later ... followon to 370 135/145 product line was the 138/148 ... which
had additional virtual memory space available. To increase the
competitiveness of 138/148 in the world market (against some foreign 370
processor clones) ... decision was to add operating system "microcode"
assist (improvement thruput of VS/1 and VM370 thruput).

I spent a lot of time with the Endicott people on measurement and
design of the vm370 feature ... old post with some of the benchmarks
used in design of the ECPS microcode assist:
http://www.garlic.com/~lynn/94.html#21 370 ECPS VM microcode assist

there was a lot of work that went into trying to make the 370/148 a
"vm370" only machine ... i.e. every 370/148 shipped would come up with
vm370 automagically (somewhat like current "LPAR" mainframes). There was
a lot of work that went into that ... before it was overruled by
corporate (as referenced in earlier post, POK was in the process
convincing corporate to kill off vm370 product). Endicott was eventually
able to salvage the vm370 product mission ... but not before the
vm370-only went by the wayside.

the 4341 was then the followon to the 370/148 ... and even tho by
that time I had switched coasts ... I still continued to do some
amount of stuff for the Endicott group.

Anne & Lynn Wheeler

unread,

Dec 14, 2009, 10:24:37 AM12/14/09

to

Anne & Lynn Wheeler <ly...@garlic.com> writes:
> this was running in regular use at cambridge a year before "real" 145
> hardware supporting virtual memory was even operational (and well before
> announce). part of the issue was that science center had a lot of users
> from various educational institutions in the boston/cambrige area
> (including students) ... and there was security concerns about 370
> virtual memory information leaking out. as a result ... the modified
> cp67 (supporting 370 virtual machine option) wasn't normally run on the
> real hardware (system used by non-employees) ... but in a 360/67 virtual
> machine. Then there was a cp67 that modified to run on "real" 370
> (instead of 360/67) ... which ran in a 370 virtual machine. Sort of
> normal operation:

re:

http://www.garlic.com/~lynn/2009r.html#33 While watching Biography about Bill Gates on CNBC last Night
http://www.garlic.com/~lynn/2009r.html#37 While watching Biography about Bill Gates on CNBC last Night

http://www.garlic.com/~lynn/2009r.html#38 While watching Biography about Bill Gates on CNBC last Night

i must have gotten labeled as (computer) security expert ... the company
got a new CSO in this period & was often the case (for large commercial
corporations), he was a former gov. employee with heavy background in
"physical" security (in this case credits included having been head of
presidential detail). i got asked to run around with the guy
... providing "computer security" information ... and some of the
"physical security" stuff rubbing off.

Anne & Lynn Wheeler

unread,

Dec 14, 2009, 12:04:03 PM12/14/09

to

Peter Flass <Peter...@Yahoo.com> writes:
> I don't believe I ever heard the 4341 call a mini, it was considered a
> small mainframe. The later 9370 was a mini, and was entirely
> rack-mountable.

re:

http://www.garlic.com/~lynn/2009r.html#33 While watching Biography about Bill Gates on CNBC last Night
http://www.garlic.com/~lynn/2009r.html#37 While watching Biography about Bill Gates on CNBC last Night
http://www.garlic.com/~lynn/2009r.html#38 While watching Biography about Bill Gates on CNBC last Night

http://www.garlic.com/~lynn/2009r.html#39 While watching Biography about Bill Gates on CNBC last Night

aka ... if numerous large corporations came and said they wanted to
place orders for several hundred (or maybe a thousand) 4341s ... but
they weren't going to be installed in traditional datacenters ... would
you turn them down??

misc. old email mentioning 4341
http://www.garlic.com/~lynn/lhwemail.html#4341

Anne & Lynn Wheeler

unread,

Dec 14, 2009, 12:28:13 PM12/14/09

to

re:
http://www.garlic.com/~lynn/2009r.html#39 While watching Biography about Bill Gates on CNBC last Night

so one weekend doing some of the data gathering runs for ECPS

http://www.garlic.com/~lynn/94.html#21 370 ECPS VM microcode assist

mentioned in this post
http://www.garlic.com/~lynn/2009r.html#39 While watching Biography about Bill Gates on CNBC last Night

i had time on 145 in the (vm370 development group) burlington mall
machine room. when the vm370 development group had outgrown space on the
3rd flr of 545 tech sq, they moved out into the vacant SBC bldg. in
burlington mall (became vacant as part of the transfer of SBC to CDC).

This was still before future system effort had been killed off
http://www.garlic.com/~lynn/submain.html#futuresys

and burlington mall being shutdown (so all the people could be
moved to pok to work on mvs/xa).

the vm370 group had some piece of future system ... as well as making
"enhancements" to vm370 for old flavor of DRM (digital rights
management) ... basically as countermeasure to ease of hardcopy
documents walking out the door (and/or copies) ... all the future system
documents were to be available in very carefully controlled environment
on vm/cms. so i came in for weekend time ... and some of the vm370 group
were very proud to point-out that the future system DRM was so secure
... "I" wouldn't eve able to get unauthorized access to the documents
... even if I was left physical alone in the machine room.

this was one of the few times I took the bait and claimed it would take
less than five minutes. I then disabled all external access to the
system in the machine room and flipped a bit in mainframe storage
(pointing out that they needed a whole additional layer of security to
prevent such compromises). basically access control was grounded on
passwords ... and I flipped a bit in the branch conditional instruction
that followed call to subroutine checking whether password was valid
(basically everything entered became a valid password).

I had to chuckle a couple decades later at the "yes card" compromise.
A chipcard payment infrastructure was defined that started to be
deployed in the 90s in europe ... and a large pilot deployment in the
US in the early part of the decade (all evidence evaporating
after "yes card" compromise became more public). old reference
to presentation at cartes 2002:
http://web.archive.org/web/20030417083810/http://www.smartcard.co.uk/resources/articles/cartes2002.html

basically trivial to create a counterfeit card. the infrastructure was
such that once the "yes card" was validated ... the card was then asked
1) if the correct PIN had been entered. a "YES CARD" was programmed to
always answer "YES". Then the "YES CARD" was asked 2) if the transaction
should be done offline ("YES") and 3) if the transaction value was
within the account credit limit ("YES") ... aka even if account number
was voided (used as anti-fraud countermeasure in magstripe
infrastructure, the transaction wouldn't go online to know that the
account was no longer valid).

related discussioin mentioning "yes card" presentations at the
ATM Integrity Task Force (same time frame as the cartes 2002
presentation):
http://www.garlic.com/~lynn/2009q.html#78 70 Years of ATM Innovation
http://www.garlic.com/~lynn/2009r.html#16 70 Years of ATM Innovation

there were other anti-fraud countermeasures that were part of the
infrastructure ... that counterfeit "YES CARD" was also programmed to
ignore ... basically a myopic focus on compromise of a valid card ...
as opposed to counterfeit "YES CARD" attacking the rest of the system.
The people doing the pilot deployment in the US had the "YES CARD"
explained to them ... and the response was to change features in valid
cards ... which had absolutely no effect on the fraud/operation of
counterfeit "YES CARD". Misc. past posts mentioning "YES CARD"
http://www.garlic.com/~lynn/subintegrity.html#yescard

Rich Alderson

unread,

Dec 14, 2009, 4:56:42 PM12/14/09

to

Andrew McLaren <m...@somewhere.com> writes:

> The DEC PDP-10 and DECSYSTEM-20 were admirable machines; but they don't
> qualify as "mainframes". They were (are?) minicomputers.

If you had any clue at all as to what you were talking about, you'd be
dangerous to someone other than yourself.

Nitwit.

--
Rich Alderson "You get what anybody gets. You get a lifetime."
ne...@alderson.users.panix.com --Death, of the Endless

Rich Alderson

unread,

Dec 14, 2009, 5:15:01 PM12/14/09

to

Jorgen Grahn <grahn...@snipabacken.se> writes:

> On Sat, 2009-12-12, Andrew McLaren wrote:
> ...

>> The perplexing bit is that some Digital folks still clamour for the
>> supposed prestige of the "mainframe" label; as though something they
>> deserved, had been unfairly withheld from them. In some circles,
>> "mainframe" is a term of abuse! Who'd want it?

> s/some/most/, surely?

No, Mr. McLaren has no idea what he's talking about.

> I haven't read much about that part of history, but it seems to me
> Digital is usually portraited as the guys who liberated companies from
> IBM serfdom by building open and cheap computers mini computers. Or
> bringing digital computing to organizations who were too small to be
> interesting to Big Blue.

And that part of the story is true, although it stretches back to 10 years
before Unix was created (on a PDP-7 18-bit system, don't forget).

But in 1964, DEC introduced the PDP-6, with up to 256K of 36-bit words,
which is to say, about the size and performance of the slightly later[1]
IBM System/360 Model 50. (Doubtless the clueless McL. will deny that that
latter machine was a mainframe, as well.) The PDP-10 is a direct follow-on
to the PDP-6, and sold into the same markets.

> Or maybe that's just my Unix bias showing.

Only in the sense that Unix has never run on the PDP-10 architecture.[2]

[1] PDP-6 announced March 1964, first ship June 1964; System/360 announced
April 1964, first ship October 1965.
[2] Ah, if only Bell Labs had bought Thompson and Ritchie that KI10 they
asked for...

Peter Flass

unread,

Dec 14, 2009, 6:39:55 PM12/14/09

to

It has a "stopwords" file of words it doesn't index, but it indexes
everything else.

Walter Bushell

unread,

Dec 14, 2009, 6:56:07 PM12/14/09

to

recent posts in this thread:
http://www.garlic.com/~lynn/2009r.html#33 While watching Biography about Bill Gates on CNBC last Night
http://www.garlic.com/~lynn/2009r.html#37 While watching Biography about Bill Gates on CNBC last Night
http://www.garlic.com/~lynn/2009r.html#38 While watching Biography about Bill Gates on CNBC last Night

http://www.garlic.com/~lynn/2009r.html#39 While watching Biography about Bill Gates on CNBC last Night

http://www.garlic.com/~lynn/2009r.html#40 While watching Biography about Bill Gates on CNBC last Night
http://www.garlic.com/~lynn/2009r.html#41 While watching Biography about Bill Gates on CNBC last Night

Dennis Ritchie

unread,

Dec 15, 2009, 9:48:06 PM12/15/09

to

"jmfbahciv" <jmfbahciv@aol> wrote in message

news:hg807...@news5.newsguy.com...
....

>> [2] Ah, if only Bell Labs had bought Thompson and Ritchie that KI10 they
>> asked for...
>>
> <grin> If they had, Unix would not be what it is today. Sometimes,
> a dearth of resources produces the best OS.

This is pretty much true for a variety of reasons. Partially our own
dearth,
but more important, the PDP-11 was an order of magnitude more
affordable for various projects first inside Bell Labs and soon in
university departments that could control their own destinies.
The later portability aspect was the third important factor.

Dennis

jmfbahciv

unread,

Dec 16, 2009, 7:22:58 AM12/16/09

to

Yup. University computer centers, who bought a PDP-10 provided
user mode access to all other departments. People in physics
labs would start to use the -10 and then "discover" they could
buy their own computer if they bought a PDP-8 or, usually,
a PDP-11. Scientists like to control their destinies. :-)

/BAH

jmfbahciv

unread,

Dec 16, 2009, 7:23:39 AM12/16/09

to

Anne & Lynn Wheeler wrote:

IME, even 20 people were too many.

/BAH

Eric Chomko

unread,

Dec 16, 2009, 2:05:10 PM12/16/09

to

On Dec 12, 9:39 am, jmfbahciv <jmfbahciv@aol> wrote:
> Andrew McLaren wrote:
> > Daddio wrote:
> >> One of the pictures they had of Bill Gates showed a DEC20 or a DEC10
> >> Mainframe computer from Digital Equipment corporation. That I used to
> >> work on for many many years.

>
> > The DEC PDP-10 and DECSYSTEM-20 were admirable machines; but they don't
> > qualify as "mainframes". They were (are?) minicomputers.
>

> WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG

Rich Alderson

unread,

Dec 16, 2009, 8:10:11 PM12/16/09

to

Rich Alderson <ne...@alderson.users.panix.com> writes:

> Andrew McLaren <m...@somewhere.com> writes:

>> The DEC PDP-10 and DECSYSTEM-20 were admirable machines; but they don't
>> qualify as "mainframes". They were (are?) minicomputers.

> If you had any clue at all as to what you were talking about, you'd be
> dangerous to someone other than yourself.

> Nitwit.

All right, I admit that I was feeling more than usually cranky when I wrote
that.

Andrew, I apologize for my rude remarks.

Rich Alderson

unread,

Dec 16, 2009, 8:47:10 PM12/16/09

to

Andrew McLaren <m...@somewhere.com> writes:

> Rich Alderson wrote:

>> No, Mr. McLaren has no idea what he's talking about.

> Ah, that's a rather inflammatory comment for a friendly discussion :-)

Not nearly so much so as my direct response, for which I've apologized
elsewhere in the thread.

> But thanks for calling me "Mr", although you're welcome to call me
> "Andrew" if you like. Let's not stand on formality.

> Okay, Rich - I admit it: I I know far more about IBM systems, and about
> DEC 16 and 32 bit machines, than I do about the 36 bit machines. That's
> true.

> So what, in your view, makes the PDP-10 qualify as a mainframe? Word
> length? Paged memory? Massbuss? Typical workloads? I'm not being
> facetious, I'm genuinely interested.

Typical workloads, for the most part, combined with large data capacities.

A KL-10 based system (DEC-10 or DEC-20) can have up to 4MW of main memory
(~20MB equivalence--not inconsiderable into the 1980s) and up to 64 500MB
disk drives (RP07) on 8 Massbus channels. Under TOPS-20 v5.x, more than
50 users can get sub-second interactive response from such a system; under
Tops-10 of the same vintage, triple that.

(With v6.x, and more with 7.0, TOPS-20 grew larger than is easily handled
in that processor, but Tops-10 is still happy as a clam on that kind of
hardware, as can be seen on our system at the museum.[1] Under those
versions of TOPS-20, you are limited to fewer than 40 users on a KL-10.)

Most people equate TOPS-20 with academic computing, and Tops-10 with "real
data processing", but Anistics, an actuarial firm in Menlo Park, had
several (4 or 5, if I remember correctly) TOPS-20 systems doing data
reductions on actuarial databases that spanned large farms of 3330-11 and
3350 disks. CitiBank ran a good portion of their business in the 1970s
and 1980s on similar systems.

I have on my own little site a scan of a 1964 magazine advert for the PDP-6,
the predecessor to the PDP-10, which does not use the word "mainframe" because
the distinction had not yet crystallized into the language, but which touts the
features which set it apart from a minicomputer and position it against the
likes of the System/360.

http://www.panix.com/~alderson/PDP-6_advert.jpg

None of this may convince you, but it is certainly why those of us who work(ed)
with these systems consider them to be mainframes rather than minis or midis or
maxis or whatever.

> FWIW I used a mainframe at work today - it was an IBM z10 EC. Not that
> that affects the argument, as such; but my conception of a mainframe is
> not just abstract (even in the historical dimension: I used IBM s/360
> long ago).

unread,

Dec 17, 2009, 7:44:24 AM12/17/09

to

No, it's not. Since you don't understand the statement, you have
no idea what a mainframe really is.

/BAH

jmfbahciv

unread,

Dec 17, 2009, 7:49:36 AM12/17/09

to

Charles Richmond wrote:
> Rich Alderson wrote:
>>
>> [snip...]
>> (With v6.x, and more with 7.0, TOPS-20 grew larger than is easily handled
>> in that processor, but Tops-10 is still happy as a clam on that kind of
>> hardware, as can be seen on our system at the museum.[1] Under those
>> versions of TOPS-20, you are limited to fewer than 40 users on a KL-10.)
>>
>
> At the university I attended, we had a DEC-20/50. It must have been set
> up in some *wimpy* way, because about 14 users would "soak" the machine.
> Fortunately, most did *not* know much about DEC equipment, and those few
> who did enjoyed using TOPS-20.

Not really. That was the number problem with TOPS-20...unable to
provide timesharing to a lot of users. If your site's workload
increased, you had to buy another -20 system to solve the problem.
With TOPS-10, you just bought another CPU and installed it.

>
>> Most people equate TOPS-20 with academic computing, and Tops-10 with
>> "real
>> data processing", but Anistics, an actuarial firm in Menlo Park, had
>> several (4 or 5, if I remember correctly) TOPS-20 systems doing data
>> reductions on actuarial databases that spanned large farms of 3330-11 and
>> 3350 disks. CitiBank ran a good portion of their business in the 1970s
>> and 1980s on similar systems.
>>
>
> A friend and I tried to get a summer job doing some kind of programming.
> We applied at this company (I think it was insurance of some kind, that
> was 30 years ago) in Dallas. This company ran COBOL on a DEC-20. The
> "hot-shot" programmer there did *not* care much for two college
> upstarts, and so we did *not* get hired.
>

he might not have had time to train you ;-).

/BAH

Charles Richmond

unread,

Dec 17, 2009, 11:42:56 AM12/17/09

to

jmfbahciv wrote:
> Charles Richmond wrote:
>> Rich Alderson wrote:
>>>
>>> [snip...] (With v6.x, and more with 7.0, TOPS-20 grew larger
>>> than is easily handled
>>> in that processor, but Tops-10 is still happy as a clam on that kind of
>>> hardware, as can be seen on our system at the museum.[1] Under those
>>> versions of TOPS-20, you are limited to fewer than 40 users on a KL-10.)
>>>
>>
>> At the university I attended, we had a DEC-20/50. It must have been
>> set up in some *wimpy* way, because about 14 users would "soak" the
>> machine. Fortunately, most did *not* know much about DEC equipment,
>> and those few who did enjoyed using TOPS-20.
>
> Not really. That was the number problem with TOPS-20...unable to
> provide timesharing to a lot of users. If your site's workload
> increased, you had to buy another -20 system to solve the problem.
> With TOPS-10, you just bought another CPU and installed it.
>

When I said the people did *not* know much about DEC equipment, I
was talking *only* about the people at the university I attended.
Some there *did* know the DEC equipment, but few enough that I had
a chance to work on the DEC-20 some.

There was a professor in the math department who was considered an
expert in numerical analysis. In order to sign this guy up, the
university guaranteed him *unlimited* computer time on the DEC-20
at night. He would have one or two jobs running on the DEC-20
overnight.

ISTR that LOTS at Stanford had *lots* of DEC-20 users. I do *not*
know about how the systems scaled, but some have told me that the
DEC-20 at our school *should* have been able to carry more
terminals. So I got the idea that the machine must have been set
up incorrectly.

>>
>>> Most people equate TOPS-20 with academic computing, and Tops-10 with
>>> "real
>>> data processing", but Anistics, an actuarial firm in Menlo Park, had
>>> several (4 or 5, if I remember correctly) TOPS-20 systems doing data
>>> reductions on actuarial databases that spanned large farms of 3330-11
>>> and
>>> 3350 disks. CitiBank ran a good portion of their business in the 1970s
>>> and 1980s on similar systems.
>>>
>>
>> A friend and I tried to get a summer job doing some kind of
>> programming. We applied at this company (I think it was insurance of
>> some kind, that was 30 years ago) in Dallas. This company ran COBOL on
>> a DEC-20. The "hot-shot" programmer there did *not* care much for two
>> college upstarts, and so we did *not* get hired.
>>
>
> he might not have had time to train you ;-).
>

We did *not* expect to be in charge. We thought the resident
programmer might find some scut work for us to do. It was probably
for the best: the boss seemed humorless and the place was full of
tobacco smoke.

Charles Richmond

unread,

Dec 17, 2009, 11:45:33 AM12/17/09

to

jmfbahciv wrote:
> Eric Chomko wrote:
>>
>> [snip...] [snip...] [snip...]

>>
>> For the benefit of the rest of us, what was the jist of it?
>
> Go look it up. PDP-10s were mainframes. Period.
>
>

Check it out on wikipedia:

http://en.wikipedia.org/wiki/PDP-10

Patrick Scheible

unread,

Dec 17, 2009, 12:19:39 PM12/17/09

to

jmfbahciv <jmfbahciv@aol> writes:

> Charles Richmond wrote:
> > Rich Alderson wrote:
> >>
> >> [snip...]
> >> (With v6.x, and more with 7.0, TOPS-20 grew larger than is easily handled
> >> in that processor, but Tops-10 is still happy as a clam on that kind of
> >> hardware, as can be seen on our system at the museum.[1] Under those
> >> versions of TOPS-20, you are limited to fewer than 40 users on a KL-10.)
> >>
> >
> > At the university I attended, we had a DEC-20/50. It must have been set
> > up in some *wimpy* way, because about 14 users would "soak" the machine.
> > Fortunately, most did *not* know much about DEC equipment, and those few
> > who did enjoyed using TOPS-20.
>
> Not really. That was the number problem with TOPS-20...unable to
> provide timesharing to a lot of users.

For high numbers of users, yes, TOPS-10 could support more than
TOPS-20. But 14 users should not have been a problem at all.
Something was wrong.

> If your site's workload
> increased, you had to buy another -20 system to solve the problem.
> With TOPS-10, you just bought another CPU and installed it.

At similar costs to buying another -20.

-- Patrick

Eric Chomko

unread,

Dec 17, 2009, 1:11:50 PM12/17/09

to

Bigger than a mini, midi and maxi, but not a supercomputer. THAT is a
mainframe. Oh, and not a cluster, like a Beowulf or some such.

Eric Chomko

unread,

Dec 17, 2009, 1:25:16 PM12/17/09

to

On Dec 17, 7:43 am, jmfbahciv <jmfbahciv@aol> wrote:
> Eric Chomko wrote:
> > On Dec 12, 9:39 am, jmfbahciv <jmfbahciv@aol> wrote:
> >> Andrew McLaren wrote:
> >>> Daddio wrote:
> >>>> One of the pictures they had of Bill Gates showed a DEC20 or a DEC10
> >>>> Mainframe computer from Digital Equipment corporation. That I used to
> >>>> work on for many many years.
> >>> The DEC PDP-10 and DECSYSTEM-20 were admirable machines; but they don't
> >>> qualify as "mainframes". They were (are?) minicomputers.
> >> WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG
>
> > I recall the term "maxi" computer from those days, a system between a
> > mini and a mainframe.
>
> Goddammit. how many WRONGs do I have to type? You are WRONG to imply
> that a PDP-10 was a maxi. And the maxi term was a joke having
> to do with hemlines.

Not where I came from! It was clearly smaller than a mainframe and
bigger than a minicomputer.

>
> >>> Arguably, the only "mainframe" produced by DEC was the VAX 9000.
> >> WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG
>
> >>> This is especially true in the late 70s, early 80s when "real"
> >>> mainframes were clearly characterised by "massive" IO channels (some of
> >>> those suckers could move 17MBps).
> >> We've discussed this before and you've still not learned.
>
> > For the benefit of the rest of us, what was the jist of it?
>
> Go look it up. PDP-10s were mainframes. Period.
>

Again, not where I came from. I used Univac 1100 series machines in
college. The only PDPs we had were PDP 11/40 and 11/45, clearly both
minicomputers.

I used a GE-600 maniframe as well in junior high school. Could a
DEC-10 do what a Univac 1100/80 or a GE-600 could do?

Eric Chomko

unread,

Dec 17, 2009, 1:26:07 PM12/17/09

to

On Dec 17, 11:45 am, Charles Richmond <friz...@tx.rr.com> wrote:
> jmfbahciv wrote:
> > Eric Chomko wrote:
>
> >> [snip...] [snip...] [snip...]
>
> >> For the benefit of the rest of us, what was the jist of it?
>
> > Go look it up. PDP-10s were mainframes. Period.
>
> Check it out on wikipedia:
>
> http://en.wikipedia.org/wiki/PDP-10
>

Yeah, I saw. Could they run 100 users simultaneously?

Patrick Scheible

unread,

Dec 17, 2009, 2:09:09 PM12/17/09

to

Eric Chomko <pne.c...@comcast.net> writes:

> On Dec 17, 7:44=A0am, jmfbahciv <jmfbahciv@aol> wrote:
> > Eric Chomko wrote:
> > > On Dec 14, 8:51 am, jmfbahciv <jmfbahciv@aol> wrote:
> > >> Anne & Lynn Wheeler wrote:
> >
> > >>> jmfbahciv <jmfbahciv@aol> writes:

> > >>>> Because it wasn't. =A0It was a mainframe and was sold as one. =A0Nob=
> ody
> > >>>> ever called it a mini. =A0We called the VAXes as Midi, a tad more
> > >>>> powerful (and bigger) than the minis. =A0VAXes didn't even run

> > >>>> as a mainframe until VMS started to transform into a genuine
> > >>>> timesharing system.
> > >>> vaxs & 43xx sold into similar/same midrange market. price and
> > >>> price/performance seem to break some threshhold and there was huge
> > >> <snip>
> >
> > >> A mainframe is a stand-alone machine unless the OS makes it perform
> > >> like a mainframe.
> >
> > > That is a circular definition.
> >

> > No, it's not. =A0Since you don't understand the statement, you have

> > no idea what a mainframe really is.
> >
>
> Bigger than a mini, midi and maxi, but not a supercomputer. THAT is a
> mainframe. Oh, and not a cluster, like a Beowulf or some such.

Just using size is a mistake. There are several other important
characteristics of mainframes: high-capacity I/O channels, high
reliability, often comparably simple calculations run on high volumes
of data.

-- Patrick

Patrick Scheible

unread,

Dec 17, 2009, 2:40:56 PM12/17/09

to

Eric Chomko <pne.c...@comcast.net> writes:

> On Dec 17, 7:43=A0am, jmfbahciv <jmfbahciv@aol> wrote:
> > Eric Chomko wrote:
> > > On Dec 12, 9:39 am, jmfbahciv <jmfbahciv@aol> wrote:
> > >> Andrew McLaren wrote:
> > >>> Daddio wrote:
> > >>>> One of the pictures they had of Bill Gates showed a DEC20 or a DEC10

> > >>>> Mainframe computer from Digital Equipment corporation. That I used t=

> o
> > >>>> work on for many many years.

> > >>> The DEC PDP-10 and DECSYSTEM-20 were admirable machines; but they don=

> 't
> > >>> qualify as "mainframes". They were (are?) minicomputers.

> > >> WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRON=

> G
> >
> > > I recall the term "maxi" computer from those days, a system between a
> > > mini and a mainframe.
> >

> > Goddammit. =A0how many WRONGs do I have to type? =A0You are WRONG to impl=
> y
> > that a PDP-10 was a maxi. =A0And the maxi term was a joke having

> > to do with hemlines.
>
> Not where I came from! It was clearly smaller than a mainframe and
> bigger than a minicomputer.
>
> >
> > >>> Arguably, the only "mainframe" produced by DEC was the VAX 9000.

> > >> WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRON=

> G
> >
> > >>> This is especially true in the late 70s, early 80s when "real"

> > >>> mainframes were clearly characterised by "massive" IO channels (some =

> of
> > >>> those suckers could move 17MBps).
> > >> We've discussed this before and you've still not learned.
> >
> > > For the benefit of the rest of us, what was the jist of it?
> >

> > Go look it up. =A0PDP-10s were mainframes. =A0Period.

> >
>
> Again, not where I came from. I used Univac 1100 series machines in
> college. The only PDPs we had were PDP 11/40 and 11/45, clearly both
> minicomputers.

PDP-11s are *not* PDP-10s. They are entirely different lines of
computers.

Are you making the assumption that a higher PDP model number means
PDP-11s are more powerful than PDP-10s? If you are, that's your first
mistake. -11s were created later, but for quite different roles.

> I used a GE-600 maniframe as well in junior high school. Could a
> DEC-10 do what a Univac 1100/80 or a GE-600 could do?

At a quick glance, the architectures look pretty similar to the
PDP-10. However there's not much about performance on wiki.

-- Patrick

Patrick Scheible

unread,

Dec 17, 2009, 2:41:32 PM12/17/09

to

Eric Chomko <pne.c...@comcast.net> writes:

> On Dec 17, 11:45=A0am, Charles Richmond <friz...@tx.rr.com> wrote:
> > jmfbahciv wrote:
> > > Eric Chomko wrote:
> >

> > >> =A0 =A0 =A0 [snip...] =A0 =A0 =A0 =A0 =A0[snip...] =A0 =A0 =A0 =A0 =A0=

> [snip...]
> >
> > >> For the benefit of the rest of us, what was the jist of it?
> >

> > > Go look it up. =A0PDP-10s were mainframes. =A0Period.

> >
> > Check it out on wikipedia:
> >
> > http://en.wikipedia.org/wiki/PDP-10
> >
>
> Yeah, I saw. Could they run 100 users simultaneously?

Yes, in the large models.

-- Patrick

Walter Bushell

unread,

Dec 17, 2009, 4:31:16 PM12/17/09

to

In article <hgckfh$fjc$1...@news.eternal-september.org>,
Charles Richmond <fri...@tx.rr.com> wrote:

> Rich Alderson wrote:
> > Peter Flass <Peter...@Yahoo.com> writes:
> >
> >> Wasn't there a -10 port of unix? I'm almost certain I saw mention of it
> >> somewhere.
> >
> > There was an attempt to port NetBSD, but it foundered on the lack of a good
> > enough C compiler.
> >
>
> There was a pretty good port of C to the Harris 800. The Harris
> 800 architecture was word addressable, each word consisting of
> three 8-bit bytes. If C could be ported well to the Harris 800,
> ISTM that a good PDP-10 implementation *could* be done.

Could use 9 bit bytes, 36 bit shorts and 72 bit longs.

--
A computer without Microsoft is like a chocolate cake without mustard.

Patrick Scheible

unread,

Dec 17, 2009, 5:54:59 PM12/17/09

to

Walter Bushell <pr...@panix.com> writes:

9-bit chars are what the PDP-10 C compilers have done. There's an
awful lot of C code written with assumption that chars are always 8
bits, though.

-- Patrick

Peter Flass

unread,

Dec 17, 2009, 5:57:36 PM12/17/09

to

Charles Richmond wrote:
>
> ISTR that LOTS at Stanford had *lots* of DEC-20 users. I do *not* know
> about how the systems scaled, but some have told me that the DEC-20 at
> our school *should* have been able to carry more terminals. So I got the
> idea that the machine must have been set up incorrectly.

Typically installations, especially colleges, tended to under-configure
the memory, which was expensive back when. Nothing slows don a TS
system like too little vcore, and all the tuning in the world can't help.

Ben Pfaff

unread,

Dec 17, 2009, 6:05:28 PM12/17/09

to

Patrick Scheible <k...@zipcon.net> writes:
> Walter Bushell <pr...@panix.com> writes:

> > Rich Alderson wrote:
> > > There was an attempt to port NetBSD, but it foundered on
> > > the lack of a good enough C compiler.

> > Could use 9 bit bytes, 36 bit shorts and 72 bit longs.
> 9-bit chars are what the PDP-10 C compilers have done. There's an
> awful lot of C code written with assumption that chars are always 8
> bits, though.

POSIX requires exactly 8-bit bytes, so surely that's a reasonable
assumption for a C program targeted at a modern Unix system.
--
Ben Pfaff
http://benpfaff.org

Anne & Lynn Wheeler

unread,

Dec 17, 2009, 6:23:53 PM12/17/09

to

Peter Flass <Peter...@Yahoo.com> writes:
> Typically installations, especially colleges, tended to
> under-configure the memory, which was expensive back when. Nothing
> slows don a TS system like too little vcore, and all the tuning in the
> world can't help.

i made page management algorithm & page I/O several times more
effective.

this showed up in comparison of the 1mbyte 360/67 cp67 system at the
grenoble science center vis-a-vis the 768kbyte 360/67 cp67 system at
cambridge science center. After fixed memory and kernel requires the CSC
768kbyte had 104 (4k) pages available and the grenoble 1mbyte had 155
(4k) pages available.

The CSC system ran 75-80 users with similar workload and response as the
grenoble system ran with 35 users.

grenoble had done a acm article about changing cp67 to reflect the
common academic virtual memory management of the time. With only 2/3rds
the virtual pages ... & not emulating the academic virtual memory
management of the era ... I was able to support twice as many users
doing effectively the same workload.

other posts in this thread:
http://www.garlic.com/~lynn/2009r.html#33 While watching Biography about Bill Gates on CNBC last Night
http://www.garlic.com/~lynn/2009r.html#37 While watching Biography about Bill Gates on CNBC last Night
http://www.garlic.com/~lynn/2009r.html#38 While watching Biography about Bill Gates on CNBC last Night
http://www.garlic.com/~lynn/2009r.html#39 While watching Biography about Bill Gates on CNBC last Night
http://www.garlic.com/~lynn/2009r.html#40 While watching Biography about Bill Gates on CNBC last Night
http://www.garlic.com/~lynn/2009r.html#41 While watching Biography about Bill Gates on CNBC last Night
http://www.garlic.com/~lynn/2009r.html#42 While watching Biography about Bill Gates on CNBC last Night

--
40+yrs virtualization experience (since Jan68), online at home since Mar1970

Patrick Scheible

unread,

Dec 17, 2009, 6:52:35 PM12/17/09

to

Ben Pfaff <b...@cs.stanford.edu> writes:

That wasn't in K&R C.

8-bit chars would be fine if C didn't also demand that word size be a
multiple of char size. All the world is not a PDP-11, VAX, or i386.

-- Patrick

Peter Flass

unread,

Dec 17, 2009, 6:48:40 PM12/17/09

to

Add, Subtract, Multiply ...

Bill Leary

unread,

Dec 17, 2009, 7:43:09 PM12/17/09

to

"Eric Chomko" <pne.c...@comcast.net> wrote in message
news:7d34dbe3-8975-4d10...@k9g2000vbl.googlegroups.com...
On Dec 12, 11:48 am, "Bill Leary" <Bill_Le...@msn.com> wrote:
> OK, Barb.
>
> Please illuminate for someone who was there and involved in buying these
> things, but didn't catch your earlier discussion.
>
> When we were shopping machines, the DEC marketroid and the engineers we
> had
> in called them "minicomputers" and occasionally "mini-mainframes." Why is
> a
> DEC PDP-10 not a minicomputer?
>

Maxi?

No. As I commented elsewhere, I took some time to research the PDP-10.
Definitely a mainframe. I think the common mistake is that people are
familiar with the other models, especially the -11, and think that the -10
was similar. Except for cosmetics, it's like they were built by two
different companies who just seemed to have a bit of common heritage about
computers.

- Bill

Morten Reistad

unread,

Dec 17, 2009, 11:10:23 PM12/17/09

to

In article <w9z8wd1...@zipcon.net>,

I have seen 115 users on a 2065 (KL10B) with 8MW of memory, three
RP06es for PS: and paging, and student registration applications
running on around 40 of the terminals.

Slowish, but fully workable. You wouldn't like to start a compile,
but a small assembly e.g. FAIL, would work.

I tuned a lot of underpowered systems in the 1980s. I always
spread out the swap with a little bit on each disk, and that
helped a lot for the underpowered systems. Like Prime 550-II's
with two megabytes of memory running 30 terminals.

-- mrr

jmfbahciv

unread,

Dec 18, 2009, 7:30:04 AM12/18/09

to

Even the most basic scut work requires training. it takes about a
year for anybody to "ramp up" w.r.t. corporate culture and internal
processes.

/BAH

jmfbahciv

unread,

Dec 18, 2009, 7:31:04 AM12/18/09

to

Charles Richmond wrote:
> jmfbahciv wrote:
>> Eric Chomko wrote:
>>>
>>> [snip...] [snip...] [snip...]
>>>
>>> For the benefit of the rest of us, what was the jist of it?
>>
>> Go look it up. PDP-10s were mainframes. Period.
>>
>>
>
> Check it out on wikipedia:
>
> http://en.wikipedia.org/wiki/PDP-10
>
>

i don't give a flying fuck what wiki says; it's often wrong. I'm
telling you how we thought and how our customers thought and
that is all that matters.

/BAH

jmfbahciv

unread,

Dec 18, 2009, 7:33:16 AM12/18/09

to

Patrick Scheible wrote:
> jmfbahciv <jmfbahciv@aol> writes:
>
>> Charles Richmond wrote:
>>> Rich Alderson wrote:
>>>> [snip...]
>>>> (With v6.x, and more with 7.0, TOPS-20 grew larger than is easily handled
>>>> in that processor, but Tops-10 is still happy as a clam on that kind of
>>>> hardware, as can be seen on our system at the museum.[1] Under those
>>>> versions of TOPS-20, you are limited to fewer than 40 users on a KL-10.)
>>>>
>>> At the university I attended, we had a DEC-20/50. It must have been set
>>> up in some *wimpy* way, because about 14 users would "soak" the machine.
>>> Fortunately, most did *not* know much about DEC equipment, and those few
>>> who did enjoyed using TOPS-20.
>> Not really. That was the number problem with TOPS-20...unable to
>> provide timesharing to a lot of users.
>
> For high numbers of users, yes, TOPS-10 could support more than
> TOPS-20. But 14 users should not have been a problem at all.
> Something was wrong.

Depends on what those 14 users were doing. If they were all
compiling the system would bog down.

>
>> If your site's workload
>> increased, you had to buy another -20 system to solve the problem.
>> With TOPS-10, you just bought another CPU and installed it.
>
> At similar costs to buying another -20.

You are thinking only of the cost of purchase. I'm talking about
the costs of operations, maintenance and foot print.

Just backing up and system software maintenance of the separate
system disks is expensive and duplicated.

/BAH

jmfbahciv

unread,

Dec 18, 2009, 7:33:59 AM12/18/09

to

Back then there were no midis nor maxis. It was joke I made.

/BAH

jmfbahciv

unread,

Dec 18, 2009, 7:36:31 AM12/18/09

to

JOBMAX at a 5-CPU installation was, IIRC, 500.
Our monitors were built with a JOBMAX of 200.

/BAH

jmfbahciv

unread,

Dec 18, 2009, 7:35:33 AM12/18/09

to

And where in the hell did I state that -11s were mainframes?
You really don't know what you're talking about.

>
> I used a GE-600 maniframe as well in junior high school. Could a
> DEC-10 do what a Univac 1100/80 or a GE-600 could do?

Babysitting junior high kids? Depends on what software was installed.

/BAH

jmfbahciv

unread,

Dec 18, 2009, 7:37:42 AM12/18/09

to

Even the single CPU systems could do that.

/BAH

jmfbahciv

unread,

Dec 18, 2009, 7:40:25 AM12/18/09

to

The PDP-10's byte instructions can handle any size byte, with a maximum
of 36 bits.

/BAH

jmfbahciv

unread,

Dec 18, 2009, 7:42:56 AM12/18/09

to

You could have the compiler do that. I can't think of any
technical reason a C and Unix couldn't be implemented. If C loves
8-bit bytes, the bit handling would be a PITA but a couple of
well-written subroutines would handle all that.

/BAH

unread,

Dec 18, 2009, 1:58:35 PM12/18/09

to

On Dec 17, 5:54 pm, Patrick Scheible <k...@zipcon.net> wrote:
> Walter Bushell <pr...@panix.com> writes:

> > In article <hgckfh$fj...@news.eternal-september.org>,

> > Charles Richmond <friz...@tx.rr.com> wrote:
>
> > > Rich Alderson wrote:

> > > > Peter Flass <Peter_Fl...@Yahoo.com> writes:
>
> > > >> Wasn't there a -10 port of unix? I'm almost certain I saw mention of it
> > > >> somewhere.
>
> > > > There was an attempt to port NetBSD, but it foundered on the lack of a good
> > > > enough C compiler.
>
> > > There was a pretty good port of C to the Harris 800. The Harris
> > > 800 architecture was word addressable, each word consisting of
> > > three 8-bit bytes. If C could be ported well to the Harris 800,
> > > ISTM that a good PDP-10 implementation *could* be done.
>
> > Could use 9 bit bytes, 36 bit shorts and 72 bit longs.
>
> 9-bit chars are what the PDP-10 C compilers have done. There's an
> awful lot of C code written with assumption that chars are always 8
> bits, though.
>

Like the Univac 1100 series, you just throw away one of the bits per
byte and use your 36 bit word as if it were a 32 bit word.

Eric Chomko

unread,

Dec 18, 2009, 2:03:35 PM12/18/09

to

Little-endian, of which they all are, does seem to be the wave of the
future. And only Intel and AMD clones of Intel at that!

I recall that Intel was so ticked at AMD for copying their chips. Well
now, if AMD didn't Intel could be written up for antitrust legistation
as a microprocessor monopoly. I guess IBM's PowerPC may save them as
well. Barely.

Morten Reistad

unread,

Dec 18, 2009, 1:34:53 PM12/18/09

to

In article <975.674T21...@kltpzyxm.invalid>,

Charlie Gibbs <cgi...@kltpzyxm.invalid> wrote:
>In article <w9z3a39...@zipcon.net>, k...@zipcon.net
>(Patrick Scheible) writes:
>
>> Eric Chomko <pne.c...@comcast.net> writes:
>>
>>> On Dec 17, 7:44=A0am, jmfbahciv <jmfbahciv@aol> wrote:
>>>
>>>> No, it's not. =A0Since you don't understand the statement, you have
>>>> no idea what a mainframe really is.
>>>
>>> Bigger than a mini, midi and maxi, but not a supercomputer. THAT is a
>>> mainframe. Oh, and not a cluster, like a Beowulf or some such.
>>
>> Just using size is a mistake. There are several other important
>> characteristics of mainframes: high-capacity I/O channels, high
>> reliability, often comparably simple calculations run on high volumes
>> of data.

The PDP10 came in several models. The latest was the KL10.

A KL10 comes in two over-width, but not too tall, large and heavy
cabinets. One of these has a scary-looking power supply inside.

There is a main processor, and a mandatory front end. There are
lots of heavy I/O cables, and these do transfers "on their own"
using DMA.

Tops20 is pretty geared towards page i/o, and let the front end
handle terminals.

The front end is a PDP11, around a foot tall inside the KL10 racks.

>My touchstones are record-oriented I/O and a batch mindset.

The PDP10 is the counterexample.

You can have timesharing even for quite large numbers of users.

But the batch subsystem, spooler and unit record system is all there,
The batch subsystem from tops20 would be a nice addition to unix-style
systems.

-- mrr

Walter Bushell

unread,

Dec 18, 2009, 2:11:40 PM12/18/09

to

In article <w9zmy1g...@zipcon.net>,
Patrick Scheible <k...@zipcon.net> wrote:

Use double words nine per.

--
A computer without Microsoft is like a chocolate cake without mustard.

Eric Chomko

unread,

Dec 18, 2009, 2:14:05 PM12/18/09

to

If it can support 100 users, then I would agree it is a mainframe. I
think that was one of the definitions from that era.

Don't mind BAH, if she met herself in a bar there would be an argument
between the two!

Eric Chomko

unread,

Dec 18, 2009, 2:16:42 PM12/18/09

to

On Dec 17, 2:40 pm, Patrick Scheible <k...@zipcon.net> wrote:
> Eric Chomko <pne.cho...@comcast.net> writes:

> > On Dec 17, 7:43=A0am, jmfbahciv <jmfbahciv@aol> wrote:
> > > Eric Chomko wrote:
> > > > On Dec 12, 9:39 am, jmfbahciv <jmfbahciv@aol> wrote:
> > > >> Andrew McLaren wrote:
> > > >>> Daddio wrote:
> > > >>>> One of the pictures they had of Bill Gates showed a DEC20 or a DEC10

> > > >>>> Mainframe computer from Digital Equipment corporation. That I used t=

> > o
> > > >>>> work on for many many years.

> > > >>> The DEC PDP-10 and DECSYSTEM-20 were admirable machines; but they don=

> > 't
> > > >>> qualify as "mainframes". They were (are?) minicomputers.

> > > >> WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRON=

> > G
>
> > > > I recall the term "maxi" computer from those days, a system between a
> > > > mini and a mainframe.
>

> > > Goddammit. =A0how many WRONGs do I have to type? =A0You are WRONG to impl=
> > y
> > > that a PDP-10 was a maxi. =A0And the maxi term was a joke having

> > > to do with hemlines.
>
> > Not where I came from! It was clearly smaller than a mainframe and
> > bigger than a minicomputer.
>
> > > >>> Arguably, the only "mainframe" produced by DEC was the VAX 9000.

> > > >> WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRONG WRON=

> > G
>
> > > >>> This is especially true in the late 70s, early 80s when "real"

> > > >>> mainframes were clearly characterised by "massive" IO channels (some =

> > of
> > > >>> those suckers could move 17MBps).
> > > >> We've discussed this before and you've still not learned.
>

> > > > For the benefit of the rest of us, what was the jist of it?
>

> > > Go look it up. =A0PDP-10s were mainframes. =A0Period.

>
> > Again, not where I came from. I used Univac 1100 series machines in
> > college. The only PDPs we had were PDP 11/40 and 11/45, clearly both
> > minicomputers.
>

> PDP-11s are *not* PDP-10s. They are entirely different lines of
> computers.

Got it. PDP-6s led to PDP-10s and PDP-8s led to PDP-11s, right? What
about the PDP-7 and 9s?

>
> Are you making the assumption that a higher PDP model number means
> PDP-11s are more powerful than PDP-10s? If you are, that's your first
> mistake. -11s were created later, but for quite different roles.

Yes, they were minis.

>
> > I used a GE-600 maniframe as well in junior high school. Could a
> > DEC-10 do what a Univac 1100/80 or a GE-600 could do?
>

> At a quick glance, the architectures look pretty similar to the
> PDP-10. However there's not much about performance on wiki.
>

Again, 100 users is a maniframe, IIRC.

Eric Chomko

unread,

Dec 18, 2009, 2:18:51 PM12/18/09

to

I lived at West Point, NY during Jr. HS and was fortunate to be able
to use the Cadet computer back in the day!

Patrick Scheible

unread,

Dec 18, 2009, 3:47:05 PM12/18/09

to

Walter Bushell <pr...@panix.com> writes:

Might work. C pointers are also expected to be multiples of char
wide, but maybe a long byte pointer with some extra bits would work.

-- Patrick

Peter Flass

unread,

Dec 18, 2009, 4:37:55 PM12/18/09

to

Unlike reasonable languages[1], C treats characters as computational
data as well as holding a character code. For example, it makes the
assumption that an unsigned char datatype can hold binary values between
0 and 255.

[1] PL/I, for example, which doesn't conflate bit, character, and
computational data.

D.J.

unread,

Dec 18, 2009, 4:51:16 PM12/18/09

to

On Fri, 18 Dec 2009 11:14:05 -0800 (PST), Eric Chomko
<pne.c...@comcast.net> wrote:
>Don't mind BAH, if she met herself in a bar there would be an argument
>between the two!

I don't know if I would buy tickets to see that, or try very
expediently to be at least a parsec away during the argument. And
probably some time after as well.

JimP.
--
Brushing aside the thorns so I can see the stars.
http://www.linuxgazette.net/ Linux Gazette
http://www.drivein-jim.net/ Drive-In movie theaters
http://poetry.drivein-jim.net/ Aug 26, 2009

des...@verizon.net

unread,

Dec 18, 2009, 5:39:27 PM12/18/09

to

Patrick Scheible <k...@zipcon.net> writes:

> Might work. C pointers are also expected to be multiples of char
> wide, but maybe a long byte pointer with some extra bits would work.

I don't recall reading anywhere that the number of bits in a char are
supposed to be an even divisor of the number of bits in a pointer.

Charlie Gibbs

unread,

Dec 18, 2009, 7:21:14 PM12/18/09

to

In article
<90adf846-1bd0-4eab...@z41g2000yqz.googlegroups.com>,
pne.c...@comcast.net (Eric Chomko) writes:

> I lived at West Point, NY during Jr. HS and was fortunate to be able
> to use the Cadet computer back in the day!

Was it an IBM 1620? ;-)

Dennis Ritchie

unread,

Dec 18, 2009, 11:40:31 PM12/18/09

to

"Peter Flass" <Peter...@Yahoo.com> wrote in message
news:hggsnl$l6u$1...@news.eternal-september.org...
....

>
> Unlike reasonable languages[1], C treats characters as computational data
> as well as holding a character code. For example, it makes the assumption
> that an unsigned char datatype can hold binary values between 0 and 255.

At least 255. Given that there was a C compiler for the PDP-10 by
1974 (MIT M.S. thesis by Alan Snyder, though I don't know how much it was
used), and also a working port of Unix for the UNIVAC 1100 series by 1984
http://cm.bell-labs.com/cm/cs/who/dmr/otherports/newp.html ,

the PDP-10 should have been straightforward if it had been desired.
The PDP-10 wouldn't even have had to cope with 1's complement
arithmetiic.

Dennis

Ahem A Rivet's Shot

unread,

Dec 19, 2009, 3:23:20 AM12/19/09

to

On Fri, 18 Dec 2009 11:14:05 -0800 (PST)

Eric Chomko <pne.c...@comcast.net> wrote:

> If it can support 100 users, then I would agree it is a mainframe. I
> think that was one of the definitions from that era.

Along with forklift access to the machine room, ...

Actually I rather thought the definition of a mainframe computer
was one that possessed a main frame holding the CPU.

--
Steve O'Hara-Smith | Directable Mirror Arrays
C:>WIN | A better way to focus the sun
The computer obeys and wins. | licences available see
You lose and Bill collects. | http://www.sohara.org/

Ahem A Rivet's Shot

unread,

Dec 19, 2009, 3:16:59 AM12/19/09

to

On 18 Dec 2009 09:31:00 -0800
Patrick Scheible <k...@zipcon.net> wrote:

> Yes, I know. It's limiting assumptions in C that are the problem. C
> has the assumption that all datatype sizes are multiples of the size
> of a char. So the would-be C implementer on PDP-10 can't use 7-bit
> chars as would be natural for ascii strings on the PDP-10, or 8-bit
> chars as a lot of the C libraries assume because the 36-bit word is
> not a multiple of 7 or 8. So 9-bit chars for C on the PDP-10 is
> generally what's done.

I don't really see a problem with 9 bit chars, a little wasteful
perhaps but otherwise not really a problem.

Ahem A Rivet's Shot

unread,

Dec 19, 2009, 3:15:34 AM12/19/09

to

On Fri, 18 Dec 2009 11:03:35 -0800 (PST)
Eric Chomko <pne.c...@comcast.net> wrote:

> I recall that Intel was so ticked at AMD for copying their chips. Well
> now, if AMD didn't Intel could be written up for antitrust legistation
> as a microprocessor monopoly. I guess IBM's PowerPC may save them as
> well. Barely.

There are an awful lot of ARMs around.

Ahem A Rivet's Shot

unread,

Dec 19, 2009, 3:18:26 AM12/19/09

to

sizeof (void *) rather implies it, sin fact sizeof pretty much
requires everything to be multiples of char size.

Peter Flass

unread,

Dec 19, 2009, 7:38:23 AM12/19/09

to

Actually, I would have thought so too. I think the reason was that the
PDP-10 was on the down slope when C was starting to become popular, so
nobody bothered.

This is just my impression, but a quick look at wikipedia shows K&R
published in 1978 and the -10 canceled in 1983, so it may in fact be the
case.

Peter Flass

unread,

Dec 19, 2009, 7:42:15 AM12/19/09

to

Ahem A Rivet's Shot wrote:
> On Fri, 18 Dec 2009 11:03:35 -0800 (PST)
> Eric Chomko <pne.c...@comcast.net> wrote:
>
>> I recall that Intel was so ticked at AMD for copying their chips. Well
>> now, if AMD didn't Intel could be written up for antitrust legistation
>> as a microprocessor monopoly. I guess IBM's PowerPC may save them as
>> well. Barely.
>
> There are an awful lot of ARMs around.
>

Depends on what you consider the relevant market for purposes of
determining a monopoly. I don't think there are many general-purpose
systems using ARM, it's mostly inmedded stuff. The power PC would now
be in the same situation except for IBM's AIX boxes, now that Apple uses
Intel.

I understand there are some ARM netbooks on the way, so that might
marginally change the situation.