Early Experiences in Computing

By Francis Townsend- : francis.townsend@bigfoot.com

I first came into intimate association with computers on November 1953. This was the day I joined 'The British Tabulating Machine Co' (BTM), which was one of the originators of what is now ICL. In those days a computer consisted of banks of thermionic valves and no way would have fitted any desk or small office. Office mechanisation was in the realms of punched cards. A punched card can be considered as holding data which, by passing through various machines, was sorted, merged with other data, results computed and final data presented to the outside world by printing on a tabulator.

Anyway back to computers. The computer being developed by BTM at that time was called The HEC, which was a first generation computer.

The HEC

A Later Version of the HEC

Like all computers it had to be programmed, but programming was then as much an art as a science. The machine basically comprised of a programme controller and arithmetic unit (the CPU) and storage in the form of a magnetic drum (the Drum).

Remember, this was before the days of solid state storage and the modern RAM. The Drum was divided into 64 tracks each consisting of 16 words. Each of the 1024 words consisted of 40 binary bits, holding either instructions or data. All programming was in the form of a binary machine code, no high level languages existed.

Execution of a program consisted of:

1. reading an instruction from the Drum into the CPU
2. executing the instruction which then consisted of:
   
   reading data from the Drum
   processing the Data
   writing results back to the Drum
   get the address of the next instruction which was contained in the present instruction
   
3. read the next instruction from the Drum.
4. repeating steps 2 and 3 until we run out of instructions

 

At all times the programmer had to remember the addresses of where data and instructions were stored. A further point to remember is that the Drum is a mechanical device rotating at a relatively slow speed so it might take up to a full revolution before the required address passed the reading heads and the required instruction/data was read. As the rotational speed of the drum was fixed it was known how long it took for each address to be read. Also it was defined how long each instruction type took to execute.

To optimise programming the machine it was ideal if, at a given instant, the programmer knew exactly where, in its rotation, the drum was. He could then ensure that instruction or data required was about to pass under the read heads. If you can imagine the surface of the Drum, it would consist of instructions and data scattered all over its surface. Instructions and data were stored in pure binary, data having been converted during input while instructions were input as binary. An instruction would consist of : Operand address, Function to be performed and the Next Instruction address.

Data and instructions were input from punched cards and final results could be either punched out on new cards or printed on the integral printer.

Apart from computers the company also manufactured electronic calculators, still based on thermionic valves.

A 541 Calculator
(Ed: I don't think this was a pocket calculator!)

These were machines that would read punched cards, (at 100 cards per minute), containing data, perform a calculation on the data and punch the answer back onto the same punched card. Data was handled in the form of 'Bini Ten' which meant that each digit was stored occupying four bits, rather like hexadecimal, but instead of a single stream of bits there were four streams running in parallel round the machine.

The Author Working on the 541

These four streams allowed handling sterling money values with 12 pennies = 1 shilling and two 10 shillings = 1 pound. The earlier versions were very simple, only allowing multiplication on an input multiplier and multiplicand arriving at a product. This was achieved with three registers, but later versions had more than three registers allowing carry over or accumulations from successive punched cards.The latest version of this machine had a magnetic drum for data storage.

The 550 Calculator Development Team	The 550 Calculator	The 555 Calculator

 

Programming of Calculators was in the form of Plug Boards.

A Plug Board was a form of patch panel where each step of a program was defined by inserting plug links. There would be a different Plug Board for each application performed and changing an application just required inserting different Plug Boards.

At this point I will dwell on a contention that existed on the definition of a computer. A computer was defined as a machine that could modify its operation based on the results of previous operations within the program. This is an ability to make a decision. In a stored program computer this can be achieved by changing an instruction, i.e. writing a modified instruction into the program. In the case of the latest calculator, where the instructions were pre-set by the Plugboard, it was impossible to change this. All that could be done was to select a different set of instructions which had been already defined. Still this is a form of decision making.

In the late 1950s transistors came on the scene which resulted in the second generation computers. The machine I was personally concerned with was known as the 1300 Series.

The 1300 Computer

I saw the prototype of this machine at a company called CDL in 1959. Steve Shirley, later to become famous as the founder of F-International, was one of the system designers. I started work on the machine in 1960 at GEC Coventry.

The 1300 was much larger, physically, than the HEC and required a very large room dedicated for it. Apart from an enlarged magnetic drum there was RAM in the form of a Core Store and magnetic tape storage which greatly increased the storage capacity. This machine also ran in the 'Bini Ten' mode. Input was by punched cards, (now running at 600 cards per minute), and paper tape was added as a later option. By modern standards the internal storage capacity was tiny. The Drum held 12,000 words. Each bank of Core Store held 400 words and there could be up to 5 banks. Magnetic tape capacity was prodigious, a reel being up to 3,600 feet long and there could be up to eight tape decks. Programming now moved into the realm of high level languages and one of the earliest available was Cobol. There were 200 of these machines built and they found their way into large companies and public utilities. The reason for the demise of this particular machine was purely technological advance and the development of integrated circuits.

The next series of computers I was concerned with was known as the 1900 Series which originated as the Ferranti FP6000 in 1963.

A 1900 Computer

This was still physically large but comprised of a number of individual units which were only connected by cables, compared with the 1300 which was bolted together as one central mass. The 1900 series used octal as the internal method of holding data. Programs were usually input in the form of paper tape whilst data was normally on punched cards. High level languages available included Cobol, Fortran and Pascal, amongst others, and programming became more of what we know now. Operating systems came into use, i.e. George III, and the machine was able to run more than one program at once using resource sharing. A standard interface was developed allowing a mix of peripherals with a minimum of program changes.

Another 1900 Computer

As time progressed successive machines were introduced, becoming physically smaller while the capacity and speed increased. The machines were still known as mainframes, as they are still known today.

(Ed: - A tear of mostalgia fell from my eyes on reading about the 1900 above. I first met the 1900 at university (1970-73), and then used one when I worked at Hawker Siddeley. I recall keeping a database of my project documentation. The dedicated programs we have today did not exist then. I wrote the database as a text file and then used the ICL line editor, under the control of George OS language, to simulate the standard add, delete and change functions. Later the company moved to an IBM 360 and trying to do the same thing with its editor and TSO operating system was a pain in comparison. The ICL and George was a magnificent computer.)

Up to 1975 I had been concerned primarily with designing, building and maintaining machines, but at this point I moved onto system design, analysis and programming. An area I stayed in until I retired in 1995.

In the late 1970s the PC as we know it came on the scene. The first machine I had was an SWTP6800. This was based on the Motorola 6800 and came in the form of a kit. For input/output I used a Teletype as no video was available. This machine had a BASIC interpreter program and was my first association with this language. Then the Commodore PET 2001, (cue fanfare), came on the scene. I still have mine be it tucked away in the garage. We all know the Commodore story from then on so I will not dwell in this area. ICL produced its first PC, be it based on the RAIR Black Box, which used the C/PM operating system.

By this time I had become a home worker, i.e. I no longer had to travel to work, only go to my office in my house. Maybe I even had the privilege of programming your Poll Tax or Rates. Programming then consisted of writing programs on coding sheets, sending the sheets to a site for punching on cards or paper tape and subsequent running on a mainframe. The result of the run was then posted back to me. It was also possible to dial-in to a mainframe using a typewriter type terminal with built in acoustic coupler. Speeds were only up to 300 baud and line interference could be a problem resulting in loss of connection.

The PC brought the power of the mainframe back to the desktop and, as they say, the rest is history.

To summarise, during my 40 plus years in computing I have seen an industry grow from a fledgeling into one that now touches all aspects of life and we could not do without it. I have seen electronics grow from large thermionic valves with their high voltages and power requirements to devices with almost unlimited capacity running on a button battery. I have travelled and been able to see the most intimate side of business and, if I had my time again, I would not change a thing.

If any readers have had associations with these machines I would like to hear from them. Also, if they would like to know more please don't hesitate to contact me. I can be reached at: francis.townsend@bigfoot.com and I have a home page, (be it simple), at: http://www.ftownsend.freeserve.co.uk

Happy computing to you all.

Journal Contents