Off Topic: But what happened to WinHEC?

NTDEV
41

In around 1991, my friend was using the latest Soviet /360 (or /370?) clone to run his Fortran scientific computations.

When he switched to 386 with 387, he said that the perf was similar, and with Weitek FPU the modern (to the date) PC was beating the old (old for US, not so old for USSR/Russia) mainframe.

throughput. A 360 equipped with a multiplexor channel

Was this something like a chain DMA-based storage controller?

performance from the bulk memory. The result was the whole machine went
down for a week

:slight_smile:

–
Maxim S. Shatskih
Windows DDK MVP
xxxxx@storagecraft.com
http://www.storagecraft.com

42

Joe,

Actually at least by the mid-1980’s the was an interpreter in the
middle. Nothing ran the
Physical instruction set except the interpreter. I was at a compiler
company that IBM approached
To take over PL/1 development for all platforms. They presented details
that did not get out
To the public, after making us essentially sign in blood. The biggest
problem was we were all
Unix types and learning IBM-ese was hard.

Don Burn
Windows Filesystem and Driver Consulting
Website: http://www.windrvr.com
Blog: http://msmvps.com/blogs/WinDrvr

“xxxxx@flounder.com” wrote in message
news:xxxxx@ntdev:

> I had a friend who worked at the IBM Rochester (Minnesota) facility, where
> the AS/400 was designed, manufactured, and programmed. He was a leader of
> one of the software teams.
>
> My understanding in talking to him is that the AS/400 was its own
> instruction set and architecture, its own operating system, and shared
> nothing with any other IBM product.
>
> In recent years, the AS/400 has used a variant of the PowerPC chips, but
> originally it was its own CPU architecture. The “assembly code” is
> actually a byte code which the original architecture executed directly,
> but now it is JIT-compiled into RISC instructions.
>
> This division was all that kept IBM alive in the early 1990s, when they
> had posted a US$5,000,000,000 loss, the largest loss any US company had
> ever posted at that point in time (it was overshadowed a few months later
> when General Motors posted a loss of US$7,500,000,000, but it turns out
> that the GM loss was a “paper loss” caused by a change in how taxes were
> calculated, so they were no longer able to write off some capital
> investments sooner than expected. IBM really lost five billion dollars).
> I was working at CMU in a research group supported by an IBM contract, and
> the contract evaporated at the end of the year, disemploying about 30
> people.
> joe
>
> >> I believe so, yes. Then again, I’ve never used an AS/400
> >
> > I saw AS/400 only at one of my former employer’s in 1994, the company was
> > trying to sell them in Russia.
> >
> > They also had an AIX box, which used PowerPC CPU and was very much like
> > Sun Sparc in everything.
> >
> > Dunno whether AS/400 used PowerPC.
> >
> > –
> > Maxim S. Shatskih
> > Windows DDK MVP
> > xxxxx@storagecraft.com
> > http://www.storagecraft.com
> >
> >
> > —
> > NTDEV is sponsored by OSR
> >
> > For our schedule of WDF, WDM, debugging and other seminars visit:
> > http://www.osr.com/seminars
> >
> > To unsubscribe, visit the List Server section of OSR Online at
> > http://www.osronline.com/page.cfm?name=ListServer
> >

43

I was an engineer/architect at IBM Rochester involved with the AS/400 design in the 1980’s. Much of the AS/400 used parts from the previous attempt at a minicomputer which was the fort knox project that sank when the operating system was so bloated it could not boot up on the hardware of the time. This was very unusual considering IBM Rochester had a reputation for being one of the finest engineering sites in the company. But we were able to reuse things like the I/O processors from fort knox which had Motorola 68K chips and PL/ compilers. The AS/400 was a massive engineering effort. You cannot believe how much we take for granted today had to be designed from scratch–hardware, firmware, OS, buses, everything. The AS/400 was announced with a sensational world wide event in the days before the internet, but the stock price of IBM went down that day. Very good business software was developed for the AS/400 and it found a strong market. But even at this time, key people in IBM were aware the glory days of proprietary hardware running proprietary operating systems with proprietary programming languages were over. Wait, what was this thread about?

44

>> Memory cost real money in those days. It was made of little magnetic

> donuts.

On IBM/360? really? (I think Soviet clones used small chips…)

Nope. I was around EC 1022 (that was year 1982) that had the whole 256 KB of core memory in a large cabinet. It had terminals with horrible vector-drawn fonts.

Later (1983), we got EC 1033 that had 2MB of static RAM in a single cabinet. Those were times. It ran a few nicer raster terminals.

45

>It turns out there was NO DOCUMENTED WAY to open a file by name from an application, or create a file by name from an application.

There actually was. You had to use assembler and make a few SVC. I’m talking about OS/360.

46

>In around 1991, my friend was using the latest Soviet /360 (or /370?) clone to run his Fortran scientific computations.

In my neck of the taiga (Tomsk), those mainframes were dismantled a few years before that. PDP-11 and LSI-11 clones replaced them.

47

My memory is that the “multiplexor channel” could handle a huge number of
simultaneous transfers, doing the equivalent of DMA from multiple
high-bandwidth devices. High-end machines could do significant fractions
of a gigabyte per second. I vaguely recollect that memory was
multiported, but that may have been one of the many custom mods our
machine had.

Gordon Bell, one of the founders of Digital Equipment Corporation (DEC)
and the architect of the PDP-11, left DEC and spent some time in the
supercomputer industry. Gordon loved to crunch numbers. He found, for
example, that the National Science Foundation grants for access to
supercomputers were typically around US$5000. With this budget, a typical
researcher could afford one minute a day of supercomputer access, and most
programs ran in about a minute. So the researcher got one result per day.
In 1990, a 386/387 system would take 24 hours to run the same program,
and this meant one result per day. So researchers were spending $3000 to
get a 386/387 that could run their problem, and with the extra $2000
getting a second, somewhat less powerful, machine for word processing,
spreadsheets, and (remember, this was 1990!) sometimes email. So they
were going after easier-to-get grants of $5000 that gave them not only
their one result per day, but other benefits their supercomputer grant
couldn’t give them.

Now compare that to a Xeon or Core series, and we can now buy, at our
local retailer, more power than the supercomputers of 1990.

The bottom line is that the supercomputer market is limited to a few, very
few, customers: weather forecasters, nuclear weapons simulators,
cryptology, and…well, there just aren’t a lot of people with big
problems and big budgets to match.
joe

In around 1991, my friend was using the latest Soviet /360 (or /370?)
clone to run his Fortran scientific computations.

When he switched to 386 with 387, he said that the perf was similar, and
with Weitek FPU the modern (to the date) PC was beating the old (old for
US, not so old for USSR/Russia) mainframe.

> throughput. A 360 equipped with a multiplexor channel

Was this something like a chain DMA-based storage controller?

> performance from the bulk memory. The result was the whole machine went
> down for a week

:slight_smile:

–
Maxim S. Shatskih
Windows DDK MVP
xxxxx@storagecraft.com
http://www.storagecraft.com

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

48

In 1968, those SVCs were considered undocumented, and we had to
reverse-engineer them by reading the OS/360 source code.

Ypu mean people did system programming in OTHER than assembler?
joe

>It turns out there was NO DOCUMENTED WAY to open a file by name from an
> application, or create a file by name from an application.

There actually was. You had to use assembler and make a few SVC. I’m
talking about OS/360.

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

49

> In my neck of the taiga (Tomsk), those mainframes were dismantled a few years before that. PDP-11

and LSI-11 clones replaced them.

From what I know from some senior guys, PDPs were never great for accounting-style apps, and had the serios drawback (compared to /360) of 64K address space, which made them not so suitable for major scientific computations.

Soviet LSI-11 (DVK-3) was my first computer. Running RT-11 OS.

Senior models of these computers had nearly the same CPU perf as 286 (with tiny RAM size though).

–
Maxim S. Shatskih
Windows DDK MVP
xxxxx@storagecraft.com
http://www.storagecraft.com

50

> Now compare that to a Xeon or Core series, and we can now buy, at our

local retailer, more power than the supercomputers of 1990.

Probably a $20 Chinese no-name MP3 player has the similar performance :slight_smile:

few, customers: weather forecasters, nuclear weapons simulators,

…semiconductor simulators, aerodynamics simulators and so on.

The more powerful the machine is, the smaller grid steps are chosen by the scientists, and so the “result per day” rule is still here :-). Nevertheless, the results are now more amazing and extend to the area which was unthinkable in 1990ies.

–
Maxim S. Shatskih
Windows DDK MVP
xxxxx@storagecraft.com
http://www.storagecraft.com

51

Probably a $20 Chinese no-name MP3 player has the similar performance
:slight_smile: $20 ??

you mean 8~10 dollors :slight_smile: i really have one with a crap 8 gb memory
usable only 1 gb :frowning: video and audio both player running which i got
for Rs 450/- about 10 $ and i am not joking i got it locally what must
be the manufacturing cost if it came all the way down to my local
market for me at 6 dollors :slight_smile: exporter cost / transporter cost /
custom duties cost / importer distributor wholeseller margin/ retailer
margin / roadside hawker margin + taxes on all above is included in 6
dollors

i infact opened it including dumping the firmaware to see why i cant
use the rest of space and in the process replaced back one of its bmp
with an edited bmp of mine :slight_smile: you can my name in this snap

and internet is really wonderful place if you search you can find
every concievable piece of information for about everything you would
ever need to lookup

http://postimage.org/image/se216hoz9/

On 5/12/12, Maxim S. Shatskih wrote:
>> Now compare that to a Xeon or Core series, and we can now buy, at our
>> local retailer, more power than the supercomputers of 1990.
>
> Probably a $20 Chinese no-name MP3 player has the similar performance :slight_smile:
>
>> few, customers: weather forecasters, nuclear weapons simulators,
>
> …semiconductor simulators, aerodynamics simulators and so on.
>
> The more powerful the machine is, the smaller grid steps are chosen by the
> scientists, and so the “result per day” rule is still here :-).
> Nevertheless, the results are now more amazing and extend to the area which
> was unthinkable in 1990ies.
>
> –
> Maxim S. Shatskih
> Windows DDK MVP
> xxxxx@storagecraft.com
> http://www.storagecraft.com
>
>
> —
> NTDEV is sponsored by OSR
>
> For our schedule of WDF, WDM, debugging and other seminars visit:
> http://www.osr.com/seminars
>
> To unsubscribe, visit the List Server section of OSR Online at
> http://www.osronline.com/page.cfm?name=ListServer
>

52

Which just reminded me that I developed a driver to connect a Sequoia
fault tolerant minicomputer to IBM mainframes using the ridiculous
blue channel-attach cables (which made fat ethernet look skinny) and
an SNA/LU6.2 protocol layer on top of that mess.

That lab had channel attach in development, token ring in development,
SNA and OSI protocol software development - a veritable graveyard of
dead tech. Surprisingly enough those Sequoia systems using motorola
68k tech are still out there chugging away, not quite dead yet as they
were actually almost impossible to kill.

Mark Roddy

On Sat, May 12, 2012 at 1:12 AM, wrote:
> My memory is that the “multiplexor channel” could handle a huge number of
> simultaneous transfers, doing the equivalent of DMA from multiple
> high-bandwidth devices. ?High-end machines could do significant fractions
> of a gigabyte per second. ?I vaguely recollect that memory was
> multiported, but that may have been one of the many custom mods our
> machine had.
>
> Gordon Bell, one of the founders of Digital Equipment Corporation (DEC)
> and the architect of the PDP-11, left DEC and spent some time in the
> supercomputer industry. ?Gordon loved to crunch numbers. ?He found, for
> example, that the National Science Foundation grants for access to
> supercomputers were typically around US$5000. ?With this budget, a typical
> researcher could afford one minute a day of supercomputer access, and most
> programs ran in about a minute. ?So the researcher got one result per day.
> ?In 1990, a 386/387 system would take 24 hours to run the same program,
> and this meant one result per day. ?So researchers were spending $3000 to
> get a 386/387 that could run their problem, and with the extra $2000
> getting a second, somewhat less powerful, machine for word processing,
> spreadsheets, and (remember, this was 1990!) sometimes email. ?So they
> were going after easier-to-get grants of $5000 that gave them not only
> their one result per day, but other benefits their supercomputer grant
> couldn’t give them.
>
> Now compare that to a Xeon or Core series, and we can now buy, at our
> local retailer, more power than the supercomputers of 1990.
>
> The bottom line is that the supercomputer market is limited to a few, very
> few, customers: weather forecasters, nuclear weapons simulators,
> cryptology, and…well, there just aren’t a lot of people with big
> problems and big budgets to match.
> ? ? ? joe
>
>> In around 1991, my friend was using the latest Soviet /360 (or /370?)
>> clone to run his Fortran scientific computations.
>>
>> When he switched to 386 with 387, he said that the perf was similar, and
>> with Weitek FPU the modern (to the date) PC was beating the old (old for
>> US, not so old for USSR/Russia) mainframe.
>>
>>> throughput. ?A 360 equipped with a multiplexor channel
>>
>> Was this something like a chain DMA-based storage controller?
>>
>>> performance from the bulk memory. ?The result was the whole machine went
>>> down for a week
>>
>> :slight_smile:
>>
>> –
>> Maxim S. Shatskih
>> Windows DDK MVP
>> xxxxx@storagecraft.com
>> http://www.storagecraft.com
>>
>>
>> —
>> NTDEV is sponsored by OSR
>>
>> For our schedule of WDF, WDM, debugging and other seminars visit:
>> http://www.osr.com/seminars
>>
>> To unsubscribe, visit the List Server section of OSR Online at
>> http://www.osronline.com/page.cfm?name=ListServer
>>
>
>
>
> —
> NTDEV is sponsored by OSR
>
> For our schedule of WDF, WDM, debugging and other seminars visit:
> http://www.osr.com/seminars
>
> To unsubscribe, visit the List Server section of OSR Online at http://www.osronline.com/page.cfm?name=ListServer

53

Ah, token ring …

Remind me why the IEEE could not agree on a bit order for the MAC layer
address? Oh never mind.

Let nothing rest as sufficient that can be overdesigned.

Cheers,
Dave Cattley

54

On 5/11/2012 7:29 PM, Calvin Guan (news) wrote:

>> //SYSUT2 DD DSN=NEWFILE,
>> // DISP=(NEW,CATLG,DELETE),
>> // SPACE=(CYL,(40,5),RLSE),
>> // DCB=(LRECL=115,BLKSIZE=1150)
>> //SYSIN DD DUMMY

Was the “Caps Lock” stuck? I feel nervous when reading all caps.
Early 6-bit terminals did not have lower-case letters.

55

My recollection is that the 370 at the computer lab in Cambridge (UK) was supplied ~1971 with 1MB of magnetic core memory. By the time I was permitted to enter the air-conditioned splendour a couple of years later they’d just got a second MB, this time using chip memory and occupying several chest-of-drawer size racks… I seem to recall an incident when an engineer dropped the EHT electrode of the operator’s terminal onto a logic rail at one point (nasty). I think the machine was called a 370/165. Was that 165ns memory cycle time? Slow though these machines were by today’s standards the sight of a punch card reader demolishing a stack of cards in seconds, or a line printer pushing out whole pages in one go gave a real sense of speed (unlike the timesharing consoles).

M

>>>>>>
----- Original Message -----
From: Maxim S. Shatskih
Newsgroups: ntdev
To: Windows System Software Devs Interest List
Sent: Friday, May 11, 2012 9:45 PM
Subject: Re:[ntdev] RE:Off Topic: But what happened to WinHEC?

Memory cost real money in those days. It was made of little magnetic
> donuts.

On IBM/360? really? (I think Soviet clones used small chips…)

And the operating system was IBM DOS

And there was also “OS” (without “D”) which was more capable, and later some VM hypervisor (MVS?).

BTW - what does “mainframe” mean? can you compare the PDP-11 which was called “mini” and IBM/360 (which was called “mainframe”) of the same time? Were /360s really much more powerful?

–
Maxim S. Shatskih
Windows DDK MVP
xxxxx@storagecraft.com
http://www.storagecraft.com

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at http://www.osronline.com/page.cfm?name=ListServer

56

The numbering system for models was intended to reflect relative power.
The /360 line ran from the 360/20 (which was an aberration, because it
wasn’t really a /360, it just had the name for marketing purposes), the
next one up was the 360/30, the smallest “real” 360. The next model up, a
standard mid-level mainframe, was the 360/50, and then the 360/40 was
introduced, which waspositioned between the 30 and 50. My recollection
was that it was oriented to COBOL-class problems, and had no
floating-point hardware. The 360/44 came along, which was a 360/40 with
floating point harware. The 360/65 was a monster mainframe, with 512KB of
main memory. The 360/75 was a “baby supercomputer” and I know of two
sites that used it. NASA used a dual-75 to simulate the Apollo missions,
and another dual-75 to do in-flight control. The calculations of how to
save Apollo-13 were run on these machines. The 360/91 was a
supercomputer, massively pipelined and would operate in what we would now
recognize as an asynchronous opportunistic execution, much like a modern
Pentium. Again, none of these had virtual memory. The first VM machine
was the 360/67, which was a 360/65 with virtual memory and an 8-slot TLB.
To the best of my recollection, none of these had caches, but I might be
wrong about the /91. But every one of them used core memory. The number
had nothing to do with anything specific about performance, just that for
any two values M1 and M2, if M2>M1, then M2 was more powerful than M1.
When the /370 line came out, the numbers were all greater than 100, so as
not to be confused with the /360 numbering system. But I stopped paying
attention to the IBM line in the early 1970s. My recollection, though, is
that VM was so successful that all models after the 360/67 (note the 75
and 91 actually predate the 67) that all the /370 and later systems had
it.
joe

My recollection is that the 370 at the computer lab in Cambridge (UK) was
supplied ~1971 with 1MB of magnetic core memory. By the time I was
permitted to enter the air-conditioned splendour a couple of years later
they’d just got a second MB, this time using chip memory and occupying
several chest-of-drawer size racks… I seem to recall an incident when an
engineer dropped the EHT electrode of the operator’s terminal onto a logic
rail at one point (nasty). I think the machine was called a 370/165. Was
that 165ns memory cycle time? Slow though these machines were by today’s
standards the sight of a punch card reader demolishing a stack of cards in
seconds, or a line printer pushing out whole pages in one go gave a real
sense of speed (unlike the timesharing consoles).

M

>>>>>>>
----- Original Message -----
From: Maxim S. Shatskih
Newsgroups: ntdev
To: Windows System Software Devs Interest List
Sent: Friday, May 11, 2012 9:45 PM
Subject: Re:[ntdev] RE:Off Topic: But what happened to WinHEC?

> Memory cost real money in those days. It was made of little magnetic
> donuts.

On IBM/360? really? (I think Soviet clones used small chips…)

> And the operating system was IBM DOS

And there was also “OS” (without “D”) which was more capable, and later
some VM hypervisor (MVS?).

BTW - what does “mainframe” mean? can you compare the PDP-11 which was
called “mini” and IBM/360 (which was called “mainframe”) of the same
time? Were /360s really much more powerful?

–
Maxim S. Shatskih
Windows DDK MVP
xxxxx@storagecraft.com
http://www.storagecraft.com

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

57

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 5/11/2012 9:30 AM, xxxxx@osr.com wrote:

//IS198CPY JOB (IS198T30500),‘COPY JOB’,CLASS=L,MSGCLASS=X //COPY01 EXEC
PGM=IEBGENER //SYSPRINT DD SYSOUT=* //SYSUT1 DD DSN=OLDFILE,DISP=SHR
//SYSUT2 DD DSN=NEWFILE, // DISP=(NEW,CATLG,DELETE), //
SPACE=(CYL,(40,5),RLSE), // DCB=(LRECL=115,BLKSIZE=1150) //SYSIN
DD DUMMY

For those of you who never had the privilege of working on an IBM 360, this
was how you copied a file. Seriously. You might think “copy OLDFILE
NEWFILE” would be enough… but NO!



The company I work for was recently nice enough to provide me a new z114, to
make our product work in that environment. Me! a guy who grew up with apple
]['s, and PC-XT’s… I’ve done a fair amount of assembly, driver driver
development, etc. But this thing was pretty much a shock.

All that JCL, its still there, and the comment earlier about running assembly
from the 1960’s is spot on. With some help from a more knowledgeable zos
person I found myself hacking assembly to create what was basically a log
roller…

So don’t get all nostalgic, some of us still have to use 3270 emulators on
our windows PC’s, and hack JCL with an editor that doesn’t do line wrap. Then
at the same time, continue to maintain modern drivers…

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iQEcBAEBAgAGBQJPsR4MAAoJEL5i86xrzcy7fL4H/3JDHvEbsmC0XiZOBPFPdiBT
+NTLx5LOmzPcKKlNQ6aAH/El7D5nQYZVjDwe81rbKqAzuMXCXBKqQgc03mcRJsEZ
RMmLG4CZUUJNb3Z5ogKcQ7pji09HmPQPhMevQuXiQ9BAVFNc8Wruod8ZvjXP4mfn
KZqIDD8SeIPI4Sfb53CKuQzFdLdf62ugQFV8TGQnelyqrY8hHImnL9UE+wYYs1jx
TQ22SXOuNhO7PzlbNkJS4cuUyzIS5+gf72/OK1C/w+rheKtmPYxQDVcHdWbY8b7c
Vb8IJJ3soMY6uyqgrGTvu8intOmO2PKW4K5z3isA5Q9XW87/+56G0iH5PDBNlPE=
=hWRH
-----END PGP SIGNATURE-----

58

My God! What did I start? Please don’t ban me for my indiscretions!

59

All you need to do to complete the circle is work C++ and linux into
the conversation.

Mark Roddy

On Tue, May 15, 2012 at 3:00 PM, wrote:
> My God! ?What did I start? ?Please don’t ban me for my indiscretions!
>
> —
> NTDEV is sponsored by OSR
>
> For our schedule of WDF, WDM, debugging and other seminars visit:
> http://www.osr.com/seminars
>
> To unsubscribe, visit the List Server section of OSR Online at http://www.osronline.com/page.cfm?name=ListServer

60

> All you need to do to complete the circle is work C++ and linux into the
conversation.

And Singularity, F#, and writing drivers in TECO.

Cheers,
Dave Cattley