How Brit computer maker beat IBM's S/360 - and Soviet spies IBM announced the System 360 in 1964 and started selling its family of modern mainframes 12 months later. But how do you outsell a problem like IBM? The S/360 cost IBM $5bn to develop, overshooting its estimated budget in 1962 by more than $4bn. Yet IBM need not have worried. Such was rate of sales for the S/360 family that …
When I worked for EDS (as a result of a take-over), I could keep my beard as long as I worked in the original company. When the contract eventually went away and EDS moved me to another location, I had to lose my beard.
I was away for a week at a time. I shaved the beard at the remote site, on Sunday night before going in to the new place Monday morning. When I got home Friday late, my wife didn't know who was running to hug her.
Now EDS is gone, and the beard is back. And my wife prefers the beard.
Win/win.
The KDF8 was effectively an RCA501.
The KDF9 was regarded as an English Electric design. The later System 4/70 and 4/75 were acknowledged internally as a successful merging of English Electric and Leo design creativity.
The System 4 range was based on the RCA Spectra 70 series which was IBM S/360 compatible . The range was envisaged as 4-10, 4.20, 4-30, 4-40, 4-50, 4-70, 4-75 - mirroring the RCA Spectra 70/15, 70/25, 70/45 range.
Different System 4 models were designed by the different merged companies. Marconi built the 4-30 - which was noticeably different and much too expensive for its market niche. The 4-50 was effectively the RCA Spectra 70/45. The System 4-70 and 4-75 used the RCA Spectra 70/55 spec with English Electric and Leo designed circuitry and logic - a model which RCA had never been able to get to market.
The delivered range was eventually pruned to 4-30, 4-50, 4-7x.
The System 4-7x series were very successful for government contracts in the UK and Commonwealth.
Like IBM S/360 these systems had many different operating systems ranging from cards, to tape, to disk.
A lot of KDF9s went to universities and sold for about £2m each. The later 4-72s were about the same price.
English Electric also had a bureau division for running customer contract work. The KDF9's 96bit double precision floating point was used for things like jet turbine blade design.
After ICL was created the bureau was spun off as a joint company with Barclays - not surprisingly called BARIC. An online TP service was run on a System 4-75 - with a 600MB disk that weighed 1.5tonnes and had water cooled bearings. It took 8 hours to archive it to tape. With its 1MB main memory it was also hired out as the largest IBM S/360 compatible in the UK - for dedication to a single computation intensive program instance for several hours. The operators imagined a customer's cost as the sound of a half-crown (12.5p) coin dropping once a second - a fortune in those days.
It was said of ICL engineers who lived for a while in the Eastern Bloc that they came to consider women with weight lifter proportions and moustaches as rather seductive. They also had horror stories of renting an apartment - only to find that the nominally resident local family of several generations were all then living packed into one adjoining room.
It was a matter of delight to the development engineers to see how fast the System 4-70 prototype was when compared to the KDF9 or even the 4-50. Tests which had taken some time to run were now over in a few seconds.
Yet even a mobile phone packs a faster cpu and more memory nowadays. The progress during the last 50 years is almost unbelievable - even when you have grown old alongside it. Each power step has gone from "fast" to "normal" to "slow" very quickly.
And nobody mentioned TSR2..
There is a flying Lightening T5 in South Africa once more, another of the last 2 seaters.
I had the joys of beginning my working life with Thorn EMI - now there's another conglomerate that made its fair share of screw ups and lost opportunities with Govt backing.
When I worked for ICL in the 70s we were leading the global fields technically.
Unfortunately the US management team who had kicked it into life left to be replaced with an old school tie British u/s management team.
ICL went rapidly downhill after I left, however I think it was more to do with the poor management than loss of me!
My first job in 1984 was a mainframe operator. We had S/360 and ICL1904 Polish made clones (R-32 and ODRA-1305). They are perfectly compatible and we were working with original Operating Systems, compilers and tools witch run without any modifications or translations.
On S/360 there were DOS/360, OS/360 MVT, HASP, Fortran and PL/1 compilers like Fortran/F, Fortran/H, PL/1 F and PL/1 Optimizing Compiler, on ICL1904 Fortran and Cobol.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
