10 April 1997                       

                         Seminar ICT/ICL 1900

 

Keeping the 1900 competitive (Virgilio Pasquali)

 

 

I will give a quick run-through of the selection and planning of the initial 1900 range, its competitive positions both at the start and at the end of its 10 years life in the market as the ICT and ICL main range and outline some of the factors that contributed to its success.

I will look at it from a Product Planning viewpoint, as I was involved with planning the 1900 during its first , and formative, six years.

 

I will start at the beginning, in 1962 in Ferranti. I was just finishing the development of Orion in West Gorton and getting ready to take the first system to Turitz in Gottemberg, to complete the development on site as it was the custom at the time.

With Atlas and Orion developments nearing completion, Hugh Devonald in Ferranti Bracknell was assembling a small team to start thinking about the design of future systems, and I was persuaded by Arthur Jackson to join that team (If I remember correctly, Derek Eldridge and Glyn Emory were the other members). I readily agreed to it, though it was taking me away from the exiting, creative and proud West Gorton environment.

When I joined the team in Bracknell in May 1993, Hugh and Derek had just recommended to Ferranti that the Ferranti-Packard FP6000 should be adopted as the basis of a Ferranti new medium system (there were other competitors, including Basic Language Machine being researched next door to us in Lily Hill by Gordon Scarrott and John Iliffe.)

Before a lot of work was done, ICT took over Ferranti, and, with the cancellation of the RCA Poplar, it was provisionally decided to adopt the 1900 as the basis for a range for the whole of ICT.

 (some technical characteristics, and how the FP6000 was changed to become the 1900- Addressing range and ICT standard interface. Remunish about Orion and the Bible)

Having been involved at the beginning, I can say with some certainty that the initial manifestation of the 1900  did not suffer from the long and careful planning of the IBM 360. We did it in a hell of a rush! The IBM 360 announcement had given us market space that we exploited by acting quickly.

Some six months after the IBM announcement we were able to announce a range of seven 1900 system models , programming aids and 18 peripherals.(Insert 1900 Announcement Chart). And we followed it with the announcement of a smaller system, the 1901, in 1965.

 

The decision to go for the 1900 range left ICT facing a considerable challenge.

ICT had to develop all its 1900 systems "in house". The external procurement route was no longer available.

ICT was at that time the number one supplier in the UK ( the UK was probably the only country in the world where IBM was not the dominant supplier) , and to maintain credibility and market share ICT had to have, and maintain competitive, a range of 1900 systems equivalent to the IBM 360 Range.

But, as we used to say proudly in those days, "IBM R&D spend is greater than ICT total revenue."

ICT had boosted its computer development skills by the acquisition of the EMI computer team and the Ferranti Computer Department. It now had skilled development groups in West Gorton and Bracknell from Atlas and Orion, and in Stevenage. But we hardly knew each other, never mind having a common development outlook and methods.

It is likely that, if we had applied today's management and risk assessment techniques to the competitive challenge we were facing then, we would probably have concluded that what we were trying to do was not possible!

 

 

 

 

 

From the performance point of view, the position in '66, after the two ranges had settled down, was that the 1900 range was competing well with the middle and lower part of the IBM 360, and, once the 1901 was announced in 1965, it was reaching below the level that the IBM 360, with its heavier architecture,  was able to reach.

 

But we had some problems at the top. The 1900 range, given parity of technology, could not compete with the top members of the IBM range in high commercial performance single systems. We wanted to be in this area of the market with its prestigious customers and its high profitability and this widely perceived shortfall provided impetus towards multiprocessor systems and eventually provided some of the drive towards a New Range.

 

Looking back, I never cease to be surprised that our engineers and programmers maintained, and improved, the competitive position of 1900 processor range vis-à-vis IBM, despite the massive difference in development resources between us and IBM.

 

Immediately after the announcement ICT set about to expand the Range, testing the limits of the Architecture both upwards and downwards.

 

As I have already mentioned, immediately after 1964 announcement we started planning the 1901. At the time conventional wisdom was that you had to have good character handling in that type of machine, and 1900 did not have any in the order code. We could not introduce character handling instructions without breaking the "upward compatibility rule". So we went for a straight 1900 machine and it was very successful. Derek Eldridge was the chairman of the working party, I was the secretary and Bill Talbot was the Stevenage development representative. He was a major force in the taking of the decision.

 

The 1901 planning was so successful that we decided to look at the feasibility of  a cheaper system, below 1901, to sell in the £40-50,000 market. We defined it in a WP and we could achieve the costs, performance and other key parameters. The proposed system was even planning to have on line enquiries via a typewriter. But the sales costs, assuming a reasonable hit rate, ruled it out on profitability and we dropped it. That established  the 1901 as the lower limit of the 1900 range, and we did not revisit that scene till the 2903 was introduced in 1973.

 

In the meantime Charlie Portman and his 1904 team (they had a head-start as they had started early from the original FP6000) improved the 1904 into 1904/5 E/F with microprogrannimg, full 1906 order code and other improvements. We even introduced a Segmentation and Paging form of addressing as an alternative, supported by the George 4 Operating System. I remember that Frank Sumner from Manchester University came to help us to specify suitable parameters.

I am sure Charlie will describe all this later. By the way, E stands for "Enhanced"; I have forgotten what F stood for but it was probably just the letter after E.

The 1904/5 E/F gave us a 1900 architecture and a template design that was used over and over without further architectural changes.

 

As I mentioned, we had a bit of a problem at the high end of the Range, where we could not make machines powerful enough to reach the top of the IBM range. To alleviate this shortfall we took two 1904/5 E/F processors and we tightly coupled them into a common store to make a "dual", the E/F, with nearly twice the power of a single 1904E/F. The Operating system was scheduling the work between the two processors, and the user could look at them as a single resource.  This "anonymous multiprocessor" design gave us valuable experience for future systems.

By then it was 1966. That was the time when there was plenty of talk about a Supercomputer: Project51, Project52, the "three wise men" and so on. Having just designed the dual 1906/7 E/F and a good Segmentation and Paging scheme, we proposed a 1908, a tightly coupled multiprocessor tying four powerful processors together into a large common store. But the proposal was overtaken by events.

 

We came out of this first phase with a well understood architecture and some stable designs.  By then we had also identified some major weaknesses in the architecture that we were unable to correct. We knew by then that, on the long term, we would need to introduce  a new range to replace the 1900.

 

The second phase of development applied advanced technologies to the well bedded and fully understood design, keeping the architecture stable and exploiting modularity.

Thus we were able, by intercepting the rapidly advancing IC technologies, to keep the 1900 Range competitive and profitable till the mid 70’s, while starting to withdraw development resources from 1969 onwards , increasingly focusing our resources 0n the “New Range”.

 

It is interesting to look back at some of the factors that I think were important (they certainly helped me) in achieving such a good result from the very limited resources that we had.

 

a)Flexible and open Architecture: The 1900 architecture, though with a number of recognised shortcomings, was very open and could be evolved quickly. The 1900 Executives were a major factor in the flexibility of the Architecture. They provided an interface between the hardware and the rest of the software, absorbing the changes and enhancements in the hardware and providing a stable compatible interface across the range in its successive manifestations. We were able to decouple the hardware and software developments. Additionally, the Executives enabled systems to be delivered without waiting for the completion of the 1900 Operating Systems, and they were a vital factor in being able to deliver the original 1900's starting in 1965, in advance of IBM. (add main hardware changes ie multis, paging, etc)

 

b) Range compatibility  was strictly adhered to. A Compatibility Committee chaired by Bruce Paterson controlled it. In principle at least all software run on all systems given adequate system resources. A number of time, during the active development life of the 1900, it was thought very desirable to change one member of the range to get a better system in that slot of the market, but it was always turned down, even if, because of that decision, the product had to be abandoned.

 

c)Modularity: The ICT Standard Interface (specified by Ron Feather at Stevenage) was a major component. It was not a new concept: Peripheral Standard Interfaces were beginning to appear in the early '60's - I came across my first experience of a peripheral standard interfaces in Ferranti when I was designing the Magnetic Tape subsystem for Orion. Peter Hall, the manager of the Computer Department, insisted that the Ampex tape decks should be interfaced so as to be interchangeable between Atlas and Orion. Most unreasonable!

But in ICT we exploited the portability that the ICT Standard Interfaces gave us very strongly as a Marketing tools and as a effort saving tool in development. Very soon other interfaces became established, like the store-processor interface for the upper subrange, that enable modular development, multiprocessors and the design of later enhancements .

 

d)Rapid application of technology:  Technology was procured externally. It forced us to look outside. We developed a good understanding of what was going on in the Industry and what the technology trends were and we had access to the most up to date technologies, that our engineers learned to apply very rapidly.

 

Besides these technical factors, some organisational factors are worth noting:

 

e)Working together: ICL united (everybody working together). When we needed to define a system, we used to call together a working party representing all interests (Planning, Marketing, Development, etc.) to tap all the expertise.  Usually a working party sat part time for two or three months . Between meetings, members were going back to their unit to consult widely with all experts (for example the development people might already be doing some pilot work on the new system and would input their initial experience). The new system under consideration was defined, and then put to the development Committee for approval. It worked well because we all knew that we only had one shot: If a development was wrongly specified, or, for some reason, failed to deliver the end system, it would leave a big hole in the range. We could not afford to have parallel competing developments for the same end system. (But West Gorton-Stevenage constructive competition kept the development up to the leading edge of technology).

The Working Party mechanism was first used to specify the 1901, immediately after the 1900 announcement. It worked so well that I continued to use that mechanism for all the new systems that I planned and defined up to 1968, when I moved out of Product Planning.

 

g)The working party mechanism was complemented by  rapid decisions by the Development Committee. The definition of the system, as developed by the appropriate working party, was costed and with a Market assessment, proposed resource allocation and costs and profitability analysis, it would be submitted to the Development Committee, the final authority. It usually got approved within one week of submission, eliminating confusion and uncertainty. When the development got approved, all I needed was a few phone calls, and the project hit the deck already running: all the members of the working party were back in senior position in their teams, fully informed and committed to project .

 

h)Focusing of development: Only the 1900 specific system components had priority in the allocation of resources. Outside procurement of all non essential subsystems (non 1900 specific) became the norm . And we turned this into an advantage, by introducing quickly “state of the art” components purchased from outside.

 

Looking at the competitive position in 1974 in terms of commercial performance, we can see that ICL (as it had then become) had improved the span of the range covered by single processor systems and was delivering  1900 systems (the S  and T series) using advanced designs with "state of the art" technology. We still priced our systems 5% below IBM on the "value for money" platform.

Besides maintaining the range competitive, we had introduced the 2903 ( a new 1900 compatible system) below the bottom of the 1900 range to compete with the IBM System 3 and we had announced and we were well advanced in the development of the three top models of the 2900 systems, the “New Range”.

 

 

 

But even after the introduction of the New Range (the 2900) in '74, the 1900 continued in the market at the small systems end of the range with the 2903/4 and eventually the ME29, while the 1900 software continued to be used in the DME versions of the 2900 as a bridging tool.

 

But from the development point of view we can look at the 1900 Range as having two different and distinct phases.

An initial phase exploring the limits of the architecture and establishing the shape of the range without major changes in technology, followed by a phase of rapid application of advanced technologies to the well bedded, stable and understood architecture.

 

 Ann.

28 Sept ‘64                                                                                                                                                                      

                       

 

                                                                1908    

                               

1906/7                                   1906/7 E/F                                                           

 

1904/5                   1904/5 E/F           

1909

                                               

1903                                                                                                                                      

1902                                                                                                                                      

                1901                                                                      

 

                                                190 1/2 

2          Arch .Dev. Phase

The 1900 weaknesses I mentioned earlier were of course only weaknesses on the long term. On the medium term the 1900 had plenty of life left, especially in the lower part of the market.

But in 1968 the merger with English Electric with its System 4 range had added a new urgency to the introduction of a much more drastically changed New Range unifying the new company. But this is another story........