Off topic: ICL 1904A and ACTP

[M J D Brown]

Peter Tomlinson certainly revived old memories for me as a programmer 
user of the ICL 1905, later upgraded to a 1907F, at West Byfleet.  In 
1968 (maybe 1969) the 1905 was brought in to replace an Elliott 503 and 
I spent a lot of time converting the old 503 machine code utility 
libraries into PLAN Assembler with the correct linkages to be called as 
external routines from ICL Algol or Fortran application programs.  The 
programming model of the 1900 series was very nice to work with, so it 
was a pleasant surprise to hear from a member of the design team.

We ran our 1900s under the GEORGE 3 operating system which allowed very 
great flexibility in simulating peripheral devices by files in its 
filestore.  Programming the disk backing store devices natively was 
rather complicated, so my practice was to pretend that all my online 
storage devices were drum memories, for which the Assembler programming 
was very simple.

I do wish I had been able to retain copies of the PLAN and GEORGE 
reference manuals.

Many thanks for reviving these memories.

Mike.

----- Original Message ----- 
From: "Peter Tomlinson" <pwt@iosis.co.uk>
To: <ukcrypto@chiark.greenend.org.uk>
Sent: Sunday, September 21, 2008 4:52 PM
Subject: Off topic: ICL 1904A and ACTP


> Although this is off topic, I hope that you will bear with me.
>
> It is 40 years ago that the ICL 1904A mainframe first burst into life.
> Conceived by ICT in 1966 and surviving across the merger to become a
> money-spinner (also in 1904X and 1904S variants), its gestation was
> greatly aided by the synchronous logic simulation system developed by
> Charles Lindsey, using the Manchester University Atlas. I well 
> remember
> a long period when every working day I coded up instruction sequences 
> as
> a colleague created them (he having unfortunately first discovered 
> that
> the so-called Large Machine Addressing Modes and Order Code were not
> exactly defined mathematically), then a young lady working for Charles
> (she an immigrant from eastern Europe, I believe - they were with us
> then) prepared a daily and highly accurate update paper tape (from my
> handwritten script) for the next update of the design description, and
> that was taken down to the Atlas computer suite for an overnight run,
> following which (if the Atlas had not crashed, or, if it did, the
> operators found time to have another 45 minute run for us) someone had
> to collect the line printer output and get it back to West Gorton for
> checking and then the daily cycle started again. We didn't simulate
> everything (so the team designing the floating point unit gave
> themselves a hard time), but later we did simulate the optional Memory
> Management Unit. And it was all put to bed by August 1968. George 4 to
> make use of the MMU took a while longer...
>
> The development of the simulation system was funded by a grant from 
> the
> govt's Advanced Computer Techniques Project.
>
> Those were the days when civil servants understood what their money 
> (our
> money) was being used for, but it didn't last much longer in a number 
> of
> fields.