Discussion:
IBM system/360 ad
(too old to reply)
h***@bbs.cpcn.com
2019-11-25 21:20:51 UTC
Permalink
https://archive.org/details/Nations-Business-1964-12/page/n51

Available with up to 8 meg of memory!

(I thought S/360 could handle up to 16 meg? But maybe in those
days no one could see needing more than 8 meg. Indeed, I think
if you wanted that much you had to get the "LCS" which was
core but slow core.)
Quadibloc
2019-11-25 23:07:31 UTC
Permalink
Post by h***@bbs.cpcn.com
https://archive.org/details/Nations-Business-1964-12/page/n51
Available with up to 8 meg of memory!
(I thought S/360 could handle up to 16 meg? But maybe in those
days no one could see needing more than 8 meg. Indeed, I think
if you wanted that much you had to get the "LCS" which was
core but slow core.)
Yes, that's right. And 16 megabytes of slow core could be attached to the
largest models in the System/360 line... which weren't available yet at the time
of that advertisement.

John Savard
Charlie Gibbs
2019-11-26 01:22:02 UTC
Permalink
Post by Quadibloc
Post by h***@bbs.cpcn.com
https://archive.org/details/Nations-Business-1964-12/page/n51
Available with up to 8 meg of memory!
(I thought S/360 could handle up to 16 meg? But maybe in those
days no one could see needing more than 8 meg. Indeed, I think
if you wanted that much you had to get the "LCS" which was
core but slow core.)
Yes, that's right. And 16 megabytes of slow core could be attached to the
largest models in the System/360 line... which weren't available yet at
the time of that advertisement.
Besides, hardly anyone would have been able to afford that much core.
I remember seeing an article in a trade rag around 1971 or so about
how IBM rocked the industry by slashing the price of a megabyte of
memory from $75,000 to a mere $15,000.

(Meanwhile, I'm slipping into my pocket a thumb drive that cost me
50 cents per gigabyte...)
--
/~\ ***@kltpzyxm.invalid (Charlie Gibbs)
\ / I'm really at ac.dekanfrus if you read it the right way.
X Top-posted messages will probably be ignored. See RFC1855.
/ \ "Alexa, define 'bugging'."
h***@bbs.cpcn.com
2019-11-27 20:54:17 UTC
Permalink
Post by Charlie Gibbs
Post by Quadibloc
Post by h***@bbs.cpcn.com
https://archive.org/details/Nations-Business-1964-12/page/n51
Available with up to 8 meg of memory!
(I thought S/360 could handle up to 16 meg? But maybe in those
days no one could see needing more than 8 meg. Indeed, I think
if you wanted that much you had to get the "LCS" which was
core but slow core.)
Yes, that's right. And 16 megabytes of slow core could be attached to the
largest models in the System/360 line... which weren't available yet at
the time of that advertisement.
Besides, hardly anyone would have been able to afford that much core.
I remember seeing an article in a trade rag around 1971 or so about
how IBM rocked the industry by slashing the price of a megabyte of
memory from $75,000 to a mere $15,000.
In those days we talked in terms of K, not meg. We were happy
to have 192K on our S/360. I think the Univac 90/30 I worked
on had 256k, which we thought was very large.

But as the 1970s wore on, technology improved and memory
got cheaper. Of course, at the same time, software got
bloated.
Post by Charlie Gibbs
(Meanwhile, I'm slipping into my pocket a thumb drive that cost me
50 cents per gigabyte...)
Yes, I got a Sandisk flash drive with 32GB for $6.
Amazing. (Though this one doesn't have the little red light
that glows when in use. I wish it did. Also, I don't know
if that is 'real' memory or compressed memory.)
Anne & Lynn Wheeler
2019-11-26 00:41:31 UTC
Permalink
Post by h***@bbs.cpcn.com
https://archive.org/details/Nations-Business-1964-12/page/n51
Available with up to 8 meg of memory!
(I thought S/360 could handle up to 16 meg? But maybe in those
days no one could see needing more than 8 meg. Indeed, I think
if you wanted that much you had to get the "LCS" which was
core but slow core.)
s/360 & s/370 architecture only having 24bit (16mbyte) addressing (real
& virtual) ... except for 360/67 which still had "real" 24bit addressing
but support 32bit virtual. Note that was the architecture ... specific
models might have much smaller number of hardware address lines
... limiting the actual amount of memory that could be attached.

by the time 3033 came around ... 16mbyte was huge constraint because of
the excessively bloated MVS kernel size as well concurrent
multiprogramming needed to try and keep system busy ... lots of page
thrashing. They did a kludge for 64mbyte support. There were two
undefined/unused bits in each page table entry (mapping virtual address
to a hardware address). The used the two unused bits to prepend to the
12bit page number (allowing 64mbytes worth of 4kbyte pages ... rather
than just 16mbytes)

Instructions (virtual and real) could only form a 24bit address ... but
a virtual address (using the page table entry hack) could map to a
64mbyte hardware address ... aka hacked 3033 that had internal 26bit
hardware address lines.

complimenting the 64mbyte page table hack ... in the original 370
architecture was fullword (31bit architecture) channel program IDALs
... which could be used to generate 64mbyte I/O transfer addresses.
--
virtualization experience starting Jan1968, online at home since Mar1970
Scott Lurndal
2019-11-26 01:31:31 UTC
Permalink
Post by Anne & Lynn Wheeler
by the time 3033 came around ... 16mbyte was huge constraint because of
the excessively bloated MVS kernel size as well concurrent
multiprogramming needed to try and keep system busy ... lots of page
thrashing. They did a kludge for 64mbyte support. There were two
undefined/unused bits in each page table entry (mapping virtual address
to a hardware address). The used the two unused bits to prepend to the
12bit page number (allowing 64mbytes worth of 4kbyte pages ... rather
than just 16mbytes)
Instructions (virtual and real) could only form a 24bit address ... but
a virtual address (using the page table entry hack) could map to a
64mbyte hardware address ... aka hacked 3033 that had internal 26bit
hardware address lines.
Intel did something similar later in the P6 days to support 36-bit physical
addresses with 32-bit virtual addresses.
h***@bbs.cpcn.com
2019-11-27 20:49:51 UTC
Permalink
Post by Anne & Lynn Wheeler
s/360 & s/370 architecture only having 24bit (16mbyte) addressing (real
& virtual) ... except for 360/67 which still had "real" 24bit addressing
but support 32bit virtual. Note that was the architecture ... specific
models might have much smaller number of hardware address lines
... limiting the actual amount of memory that could be attached.
Anyone know how much real memory does the Z series support today?

Unlike the past, it's hard to get firm information from IBM
on various models, even if they even have specific models.
Andy Burns
2019-11-27 21:23:00 UTC
Permalink
Post by h***@bbs.cpcn.com
Anyone know how much real memory does the Z series support today?
Unlike the past, it's hard to get firm information from IBM
on various models, even if they even have specific models.
40TB

<https://www.mainline.com/ibm-z15-september-12-2019-announcement>

since that's in RAIM config, I assume less useable memory if you
actually mirror/stripe/whatever it?
h***@bbs.cpcn.com
2019-11-27 21:39:25 UTC
Permalink
Post by Andy Burns
Post by h***@bbs.cpcn.com
Anyone know how much real memory does the Z series support today?
Unlike the past, it's hard to get firm information from IBM
on various models, even if they even have specific models.
40TB
<https://www.mainline.com/ibm-z15-september-12-2019-announcement>
since that's in RAIM config, I assume less useable memory if you
actually mirror/stripe/whatever it?
OMG.

But I must admit caching is nice. First time I read a file
it takes a minute or two. But each successive access
runs very fast. I assumed the file is stored in memory
in its entirety.
J. Clarke
2019-11-27 23:18:43 UTC
Permalink
Post by h***@bbs.cpcn.com
Post by Andy Burns
Post by h***@bbs.cpcn.com
Anyone know how much real memory does the Z series support today?
Unlike the past, it's hard to get firm information from IBM
on various models, even if they even have specific models.
40TB
<https://www.mainline.com/ibm-z15-september-12-2019-announcement>
since that's in RAIM config, I assume less useable memory if you
actually mirror/stripe/whatever it?
OMG.
But I must admit caching is nice. First time I read a file
it takes a minute or two. But each successive access
runs very fast. I assumed the file is stored in memory
in its entirety.
I did an experiment many years ago. I had two machines. One was a 1
GHz Pentium and the other was a 450 MHz Xeon. There was a particular
dataset and program that I was working with. I noticed that the
dataset was larger than the cache on the Pentium but smaller than the
cache on the Xeon. Same code, same OS, same as much else as possible,
the Xeon ran that particular program on that particular dataset twice
as fast.

David Wade
2019-11-26 21:35:29 UTC
Permalink
Post by h***@bbs.cpcn.com
https://archive.org/details/Nations-Business-1964-12/page/n51
Available with up to 8 meg of memory!
(I thought S/360 could handle up to 16 meg? But maybe in those
days no one could see needing more than 8 meg. Indeed, I think
if you wanted that much you had to get the "LCS" which was
core but slow core.)
The ARCHITECTURE could support up to 16Mb and indeed the 360/67 could
support more, but physically, most 360's could only interface to smaller
amounts of RAM and only implemented a subset of the address lines.
In addition if it spread out too far the propogation delays will slow
the machine down.

For example a 360/40 could have up to 256k and I think a 360/67 could go
up to 2048k. There is a picture of a 360/67 with 512k here:-

http://history.cs.ncl.ac.uk/anniversaries/40th/images/ibm360_672/slide07.html

The core is in the four cabinet right at the back. When I used that
machine it had been upgraded to 1024K so had 8 of the big double cabinets.

The functional characteristics manuals which cover the specs are here:-

http://www.bitsavers.org/pdf/ibm/360/funcChar/

other machines have similar restrictions,so the early VAXs could only
handle 1 or 2Mb of memory, my VLC has 24Mb.

Dave
Bob Eager
2019-11-26 22:21:58 UTC
Permalink
Post by David Wade
The ARCHITECTURE could support up to 16Mb and indeed the 360/67 could
support more, but physically, most 360's could only interface to smaller
amounts of RAM and only implemented a subset of the address lines.
In addition if it spread out too far the propogation delays will slow
the machine down.
other machines have similar restrictions,so the early VAXs could only
handle 1 or 2Mb of memory, my VLC has 24Mb.
Also see the 80386SX; only 24 address lines but full 32 bit architecture.
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
Charlie Gibbs
2019-11-26 22:50:46 UTC
Permalink
Post by David Wade
Post by h***@bbs.cpcn.com
https://archive.org/details/Nations-Business-1964-12/page/n51
Available with up to 8 meg of memory!
(I thought S/360 could handle up to 16 meg? But maybe in those
days no one could see needing more than 8 meg. Indeed, I think
if you wanted that much you had to get the "LCS" which was
core but slow core.)
The ARCHITECTURE could support up to 16Mb and indeed the 360/67 could
support more, but physically, most 360's could only interface to smaller
amounts of RAM and only implemented a subset of the address lines.
In addition if it spread out too far the propogation delays will slow
the machine down.
For example a 360/40 could have up to 256k and I think a 360/67 could go
up to 2048k. There is a picture of a 360/67 with 512k here:-
http://history.cs.ncl.ac.uk/anniversaries/40th/images/ibm360_672/slide07.html
The core is in the four cabinet right at the back. When I used that
machine it had been upgraded to 1024K so had 8 of the big double cabinets.
The functional characteristics manuals which cover the specs are here:-
http://www.bitsavers.org/pdf/ibm/360/funcChar/
Yours must have been an older machine. The 360/67 we had at UBC
had 256K in each cabinet. The above manual describes this as two
128K units per cabinet.

Although propagation delays were a factor, I suspect that marketing
was as well. Some shops had to go to a larger CPU than they needed
in order to get enough memory. Third-party suppliers were quick to
jump in. I once used a 360/30 that had 128K of memory, even though
you could only go to 64K on a stock machine. A switch and indicator
light for the extra address bit was placed in an unused portion of
the panel.

I once read about how Greyhound Computer Corporation would buy
360/30s that came off lease and add memory to them. Apparently
you could get up to 512K if you wanted.
--
/~\ ***@kltpzyxm.invalid (Charlie Gibbs)
\ / I'm really at ac.dekanfrus if you read it the right way.
X Top-posted messages will probably be ignored. See RFC1855.
/ \ "Alexa, define 'bugging'."
David Wade
2019-11-27 00:41:04 UTC
Permalink
Post by Charlie Gibbs
Post by David Wade
Post by h***@bbs.cpcn.com
https://archive.org/details/Nations-Business-1964-12/page/n51
Available with up to 8 meg of memory!
(I thought S/360 could handle up to 16 meg? But maybe in those
days no one could see needing more than 8 meg. Indeed, I think
if you wanted that much you had to get the "LCS" which was
core but slow core.)
The ARCHITECTURE could support up to 16Mb and indeed the 360/67 could
support more, but physically, most 360's could only interface to smaller
amounts of RAM and only implemented a subset of the address lines.
In addition if it spread out too far the propogation delays will slow
the machine down.
For example a 360/40 could have up to 256k and I think a 360/67 could go
up to 2048k. There is a picture of a 360/67 with 512k here:-
http://history.cs.ncl.ac.uk/anniversaries/40th/images/ibm360_672/slide07.html
The core is in the four cabinet right at the back. When I used that
machine it had been upgraded to 1024K so had 8 of the big double cabinets.
The functional characteristics manuals which cover the specs are here:-
http://www.bitsavers.org/pdf/ibm/360/funcChar/
Yours must have been an older machine. The 360/67 we had at UBC
had 256K in each cabinet. The above manual describes this as two
128K units per cabinet.
Not sure we are talking about the same cabinet. The store came in 256k
chunks, but when you actually checked there were 2 x 128k cabinets.
I assume it was interleaved...
Post by Charlie Gibbs
Although propagation delays were a factor, I suspect that marketing
was as well. Some shops had to go to a larger CPU than they needed
in order to get enough memory. Third-party suppliers were quick to
jump in. I once used a 360/30 that had 128K of memory, even though
you could only go to 64K on a stock machine. A switch and indicator
light for the extra address bit was placed in an unused portion of
the panel.
I seem to remember hearing that some of those upgrades needed a lot of
re-working of the hardware, on the other hand that was the time of the
expensive "no change" upgrade. Printers where the speed was set by a
link, but getting it changed was expensive.

Disk drives that were 100mb until a wire was cut when they became 200Mb...

etc. etc. etc.
Post by Charlie Gibbs
I once read about how Greyhound Computer Corporation would buy
360/30s that came off lease and add memory to them. Apparently
you could get up to 512K if you wanted.
That was a lot of memory for a 30....

Dave
Charlie Gibbs
2019-11-27 05:49:36 UTC
Permalink
Post by David Wade
Post by Charlie Gibbs
Although propagation delays were a factor, I suspect that marketing
was as well. Some shops had to go to a larger CPU than they needed
in order to get enough memory. Third-party suppliers were quick to
jump in. I once used a 360/30 that had 128K of memory, even though
you could only go to 64K on a stock machine. A switch and indicator
light for the extra address bit was placed in an unused portion of
the panel.
I seem to remember hearing that some of those upgrades needed a lot of
re-working of the hardware, on the other hand that was the time of the
expensive "no change" upgrade. Printers where the speed was set by a
link, but getting it changed was expensive.
Disk drives that were 100mb until a wire was cut when they became 200Mb...
etc. etc. etc.
This particular /30 also had 3rd party disk and tape drives, along
with the memory upgrades. IBM was Not Pleased.

The ultimate 3rd party story I heard was the shop that had a 370/168
with all plug-compatible peripherals - and then they replaced the
processor with an Amdahl. The only IBM part left was the software.
--
/~\ ***@kltpzyxm.invalid (Charlie Gibbs)
\ / I'm really at ac.dekanfrus if you read it the right way.
X Top-posted messages will probably be ignored. See RFC1855.
/ \ "Alexa, define 'bugging'."
David Wade
2019-11-27 11:22:30 UTC
Permalink
Post by Charlie Gibbs
Post by David Wade
Post by Charlie Gibbs
Although propagation delays were a factor, I suspect that marketing
was as well. Some shops had to go to a larger CPU than they needed
in order to get enough memory. Third-party suppliers were quick to
jump in. I once used a 360/30 that had 128K of memory, even though
you could only go to 64K on a stock machine. A switch and indicator
light for the extra address bit was placed in an unused portion of
the panel.
I seem to remember hearing that some of those upgrades needed a lot of
re-working of the hardware, on the other hand that was the time of the
expensive "no change" upgrade. Printers where the speed was set by a
link, but getting it changed was expensive.
Disk drives that were 100mb until a wire was cut when they became 200Mb...
etc. etc. etc.
This particular /30 also had 3rd party disk and tape drives, along
with the memory upgrades. IBM was Not Pleased.
The ultimate 3rd party story I heard was the shop that had a 370/168
with all plug-compatible peripherals - and then they replaced the
processor with an Amdahl. The only IBM part left was the software.
I believe that at one time NERC had 3rd party cartridge drives so 3480?
on all its IBM boxes, but the VAXs had IBM drives....

Dave
Charlie Gibbs
2019-11-27 19:54:26 UTC
Permalink
Post by David Wade
Post by Charlie Gibbs
The ultimate 3rd party story I heard was the shop that had a 370/168
with all plug-compatible peripherals - and then they replaced the
processor with an Amdahl. The only IBM part left was the software.
I believe that at one time NERC had 3rd party cartridge drives so 3480?
on all its IBM boxes, but the VAXs had IBM drives....
When I was with Univac I visited a customer site to do some work on their
90/30. While waiting for a run I wandered over to the 11/70 that was in
the same room, and popped the back off one of the RP06 drives, only to
find a plate reading "ISS / Sperry Univac".
--
/~\ ***@kltpzyxm.invalid (Charlie Gibbs)
\ / I'm really at ac.dekanfrus if you read it the right way.
X Top-posted messages will probably be ignored. See RFC1855.
/ \ "Alexa, define 'bugging'."
h***@bbs.cpcn.com
2019-11-27 21:05:55 UTC
Permalink
Post by Charlie Gibbs
When I was with Univac I visited a customer site to do some work on their
90/30. While waiting for a run I wandered over to the 11/70 that was in
the same room, and popped the back off one of the RP06 drives, only to
find a plate reading "ISS / Sperry Univac".
Our 90/30 had circuit cards by Intel.

Bell Telephone bought a lot of hardware from their rival,
Automatic Electric Co.
h***@bbs.cpcn.com
2019-11-27 21:04:14 UTC
Permalink
Post by Charlie Gibbs
This particular /30 also had 3rd party disk and tape drives, along
with the memory upgrades. IBM was Not Pleased.
The S/360 history explains some of what was going on. Part
of the third-party hardware was the result of IBM engineers
'jumping ship' after S/360 development, taking their
expertise to newly formed companies. IBM was pissed about
that. But my personal impression is that IBM somewhat created
that situation by working their people ridiculously hard
during S/360 days and then not compensating them adequately.

Another part was that S/360 was very successful and IBM
was unable to meet demand.

Another part was that IBM wasn't cheap. IBM gave full
support, but at a price. Later, after unbundling, IBM
was more competitive.

Of course, a third party company was exploiting IBM's
research and development. But there were risks--they
could make a lot of money, but if IBM upgraded its
hardware, they could lose a lot of money.

Tom Watson Jr initially had the attitude that IBM
was entitled to own virtually the entire DP marketplace.
He enacted policies got IBM sued. Later he woke up
and changed their policies, so when the govt sued, IBM
won.
Post by Charlie Gibbs
The ultimate 3rd party story I heard was the shop that had a 370/168
with all plug-compatible peripherals - and then they replaced the
processor with an Amdahl. The only IBM part left was the software.
Not unusual.

Our hospital leased its 360 from a third party leasing company.
IBM hated that, but they still gave us nice support. We rented
a few things and maintenance from them.
Anne & Lynn Wheeler
2019-11-27 23:11:56 UTC
Permalink
Post by h***@bbs.cpcn.com
Another part was that S/360 was very successful and IBM
was unable to meet demand.
Another part was that IBM wasn't cheap. IBM gave full
support, but at a price. Later, after unbundling, IBM
was more competitive.
Of course, a third party company was exploiting IBM's
research and development. But there were risks--they
could make a lot of money, but if IBM upgraded its
hardware, they could lose a lot of money.
Tom Watson Jr initially had the attitude that IBM
was entitled to own virtually the entire DP marketplace.
He enacted policies got IBM sued. Later he woke up
and changed their policies, so when the govt sued, IBM
won.
rise and fall of ibm
https://www.ecole.org/en/session/49-the-rise-and-fall-of-ibm
article ... mention coutmeasure to clone controllers&devices
https://www.ecole.org/en/65/CM200195-ENG.pdf
IBM tried to react by launching a major project called the 'Future
System' (FS) in the early 1970's. The idea was to get so far ahead
that the competition would never be able to keep up, and to have such
a high level of integration that it would be impossible for
competitors to follow a compatible niche strategy. However, the
project failed because the objectives were too ambitious for the
available technology. Many of the ideas that were developed were
nevertheless adapted for later generations. Once IBM had acknowledged
this failure, it launched its 'box strategy', which called for
competitiveness with all the different types of compatible
sub-systems. But this proved to be difficult because of IBM's cost
structure and its R&D spending, and the strategy only resulted in a
partial narrowing of the price gap between IBM and its rivals.

End of ACS/360, IBM executives worried that it would advance the
state-of-art too fast and they would loose control of the market. Amdahl
leaves IBM shortly afterwards and starts his own clone processor
company.
https://people.cs.clemson.edu/~mark/acs_end.html

Note that internal politics during the FS period was stopping/shutting
down 370 efforts (because FS was completely different and was going to
completely replace 370). The lack of new 370 products during the FS
period is credited with giving clone processor makers market foothold.

23jun1969 unbundling posts
http://www.garlic.com/~lynn/submain.html#unbundling
future system posts
http://www.garlic.com/~lynn/submain.html#futuresys

clone controller trrivia: 3 people from science center installed
(virutal machine) CP67 at the univ. last weekend jan1968.
https://en.wikipedia.org/wiki/CP/CMS

It had 2741&1052 terminal support with automatic terminal type
identification. The univ. had some number of ascii/tty terminals and so
I extended the support to ascii/tty (including auto-terminal type
identification, using the IBM terminal controller SAD CCW to switch port
terminal type scanner). I then wanted to extend to having a single
dial-up number of all terminal types (hunt group) ... but it didn't
quiet work, while SAD command allowed switching scanner type, all ports
had fixed hardwired port speed (could dynamically connect terminals to
any port with wrong speed).

This was motivation for the univ. to start clone controller project,
building controler channel attach board for Interdata/3 programmed to
emulate IBM terminal controller ... but including support for doing
dynamic line speed. This was then enhanced with a Interdata/4 for the
channel interface and cluster of Interdata/3s for port scanners ... and
Interdata started selling them as IBM clone controllers (later under
Perkin/Elmer logo after they bought Interdata). Four of us get written
up responsible for (some part of) clone controller business
https://en.wikipedia.org/wiki/Interdata

science center posts
http://www.garlic.com/~lynn/subtopic.html#545tech
clone controller posts
http://www.garlic.com/~lynn/subtopic.html#360pcm

tty support trivia ... I did hack using 1byte field for tty/ascii
terminal lengths. IBM had included the support in standard distributed
system. One of the cp67 installations in bldg. across the court from 545
in tech sq ... somebody got a ascii plotter down at harvard, needing
1200? byte line length. The just change max length field to 1200, but
not the rest of the code ... so the system would crash as soon as tty
connect.
http://www.multicians.org/thvv/360-67.html
--
virtualization experience starting Jan1968, online at home since Mar1970
Loading...