Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

What Banking Will Be Like - 1969

9 views
Skip to first unread message

Peter Grange

unread,
May 18, 2012, 10:38:41 AM5/18/12
to
Don't know if this has been posted before - apologies if so.

Clip from BBC "Tomorrow's World" in 1969. Can't place the hardware,
but what's coming out of the printer looks suspiciously like a crash
dump...
http://www.retronaut.co/2011/08/new-banking-tomorrows-world-1969/

--

Peter Grange

Joe Morris

unread,
May 18, 2012, 10:38:30 PM5/18/12
to
Dump it is. The data lines are hex, and I was able to make out a header
"LOAD MODULE TEMP" (I'm not completely sure of "temp"). There's an EBCDIC
string ("ILINIT") and all sorts of machine instructions that I could somehow
read despite having left the mainframe world almost 20 years ago.

Joe


Ken Wheatley

unread,
May 19, 2012, 5:28:01 AM5/19/12
to
Is it Burroughs?

Peter Grange

unread,
May 19, 2012, 7:53:48 AM5/19/12
to
On Sat, 19 May 2012 10:28:01 +0100, Ken Wheatley <k...@birchanger.com>
wrote:
That would have been my guess, Burroughs had a lot of success in the
banks in the late 60s.

Louis Krupp

unread,
May 19, 2012, 8:25:30 AM5/19/12
to
On Sat, 19 May 2012 10:28:01 +0100, Ken Wheatley <k...@birchanger.com>
wrote:

An early pre-release B6500, maybe. I know Burroughs was still doing
cross-compliations of the MCP in 1969. The Conrac (?) displays look
familiar. If Joe says the dump contains instructions, then it
probably does. Hex A7 is a B6500 RETN (I looked it up), hex 92 is an
LNOT. The Mark Stack (AE) ... Enter (AB) sequence would be a
give-away, but at the moment, I don't see any of those.

Louis


Louis

Peter Flass

unread,
May 19, 2012, 8:51:10 AM5/19/12
to
That was because you had an instruction set that was simple and
rational, unlike today's hodpodges.


--
Pete

Tim Shoppa

unread,
May 19, 2012, 10:44:31 AM5/19/12
to
I like the model-computer-room at the very beginning. They pretty much
nailed what a large company's mainframe system's disk farm would look
like 20+ years later.

Tim.

Stephen Wolstenholme

unread,
May 19, 2012, 11:54:32 AM5/19/12
to
On Sat, 19 May 2012 10:28:01 +0100, Ken Wheatley <k...@birchanger.com>
wrote:

It looks like a ICL System 4 that I worked on.

Steve

--
Neural Network Software. http://www.npsl1.com
EasyNN-plus. Neural Networks plus. http://www.easynn.com
SwingNN. Forecast with Neural Networks. http://www.swingnn.com
JustNN. Just Neural Networks. http://www.justnn.com

Joe Morris

unread,
May 19, 2012, 12:48:38 PM5/19/12
to
Looking at the dump, I'll call it certain that it's a dump of an S/360 (or
compatible machine) program. We see only about 11 bytes of data on each
line, but some examples:

2C722: MVC B1B(A,8),?C8(A) /move 8 characters
B 0B?(A) /unconditional branch
2C740: BNZ (or BNE) 080(A) /branch on nonzero or unequal
MVC B0B(A,8),BCB(A)
B ???
2C760: CLI 818(A),X'00' /compare one byte of memory
BE 0FA(A) /branch if equal
LA 1,31?(A) /load address to register
2C782 has the character string I quoted - in EBCDIC, not ASCII
2C7A4: XC 808(A,8),808(A) /exclusive OR 8 bytes memory-to-memory
BZ ???(?)
2C7C0: XR 4,4 /exclusive OR register-to-register

Fragments, but each one is self-consistent, and all of the base registers
refer to R10. (Using exclusive OR was a common way of testing a register or
field for zero or nonzero status.)

Joe


Dan Espen

unread,
May 19, 2012, 2:33:12 PM5/19/12
to
"Joe Morris" <j.c.m...@verizon.net> writes:

> "Louis Krupp" <lkr...@nospam.pssw.com.invalid> wrote:
>> Ken Wheatley <k...@birchanger.com> wrote:
>>>On 2012-05-18 14:38:41 +0000, Peter Grange said:
>
>>>> Clip from BBC "Tomorrow's World" in 1969. Can't place the hardware,
>>>> but what's coming out of the printer looks suspiciously like a crash
>>>> dump...
>>>> http://www.retronaut.co/2011/08/new-banking-tomorrows-world-1969/
>
>>>Is it Burroughs?
>
>> An early pre-release B6500, maybe. I know Burroughs was still doing
>> cross-compliations of the MCP in 1969. The Conrac (?) displays look
>> familiar. If Joe says the dump contains instructions, then it
>> probably does. Hex A7 is a B6500 RETN (I looked it up), hex 92 is an
>> LNOT. The Mark Stack (AE) ... Enter (AB) sequence would be a
>> give-away, but at the moment, I don't see any of those.
>
> Looking at the dump, I'll call it certain that it's a dump of an S/360 (or
> compatible machine) program. We see only about 11 bytes of data on each
> line, but some examples:
>
> 2C722: MVC B1B(A,8),?C8(A) /move 8 characters

Move 8 would be:

MVC B1B(8,R10),C8(R10)

Of course if you see:

D208AB1BA0C8
^^

That's move 9 characters.


> B 0B?(A) /unconditional branch
> 2C740: BNZ (or BNE) 080(A) /branch on nonzero or unequal
> MVC B0B(A,8),BCB(A)
> B ???
> 2C760: CLI 818(A),X'00' /compare one byte of memory
> BE 0FA(A) /branch if equal
> LA 1,31?(A) /load address to register
> 2C782 has the character string I quoted - in EBCDIC, not ASCII
> 2C7A4: XC 808(A,8),808(A) /exclusive OR 8 bytes memory-to-memory

Same here.

> BZ ???(?)
> 2C7C0: XR 4,4 /exclusive OR register-to-register
>
> Fragments, but each one is self-consistent, and all of the base registers
> refer to R10. (Using exclusive OR was a common way of testing a register or
> field for zero or nonzero status.)

The above XR clears R4 to zero.
The operation sets the condition code but there's no "test" involved,
the result is always zero.

--
Dan Espen

Joe Morris

unread,
May 19, 2012, 7:05:16 PM5/19/12
to
"Dan Espen" <des...@verizon.net> wrote:
> "Joe Morris" <j.c.m...@verizon.net> writes:

>> 2C722: MVC B1B(A,8),?C8(A) /move 8 characters
>
> Move 8 would be:
>
> MVC B1B(8,R10),C8(R10)

Yup (and ditto on the XR). It's been a while since I wrote 360 code.

Joe


Quadibloc

unread,
May 22, 2012, 2:55:02 PM5/22/12
to
On May 19, 9:54 am, Stephen Wolstenholme <st...@npsl1.com> wrote:

> It looks like a ICL System 4 that I worked on.

The terminals resemble some made by Burroughs, so that covers the
people with that reply.

Even before seeing the part of the thread which included disassembly,
as soon as someone spoke of a "simple, rational" instruction set easy
to read from a hex dump I thought of the System/360. As it happens,
the ICL System 4 was based on the RCA Spectra 70, and many of us will
recall that the Spectra 70 had the same problem state instruction set
as the 360, although its privileged instructions were different.

So everybody gets to be right!

John Savard

Stephen Wolstenholme

unread,
May 23, 2012, 4:55:11 AM5/23/12
to
On Tue, 22 May 2012 11:55:02 -0700 (PDT), Quadibloc
<jsa...@ecn.ab.ca> wrote:

>Even before seeing the part of the thread which included disassembly,
>as soon as someone spoke of a "simple, rational" instruction set easy
>to read from a hex dump I thought of the System/360. As it happens,
>the ICL System 4 was based on the RCA Spectra 70, and many of us will
>recall that the Spectra 70 had the same problem state instruction set
>as the 360, although its privileged instructions were different.


The English Electric 4/50 was based on Spectra 70. The much more
powerful 4/70 was the series I worked on. It was also developed by
English Electric and it later merged into ICL. The 4/70 used the same
instruction set as the IBM 360. I can't remember the privilege
instructions but, as you say, they were different to the 360. I
remember there were multiple levels of privilege from processors up to
user level software.
0 new messages