The RCA 301

Well, ok. It was not really my first computer. It was my third.

Core memory doubled to 40,000 characters! Magnetic tape capability increased to twelve or more 66,000 character/second tape units !

This was state-of-the-art in 1962 ! Memory looked like this – tiny ferrous iron-core doughnuts. Each core had 3 wires thru the center, one for positive charges, one for negative charges and a ‘read’ wire to sense the magnetic state of the core. One computer operator in our unit became unhappy about a missed promotion. One evening, he placed a jelly doughnut on the top of core memory in our RCA501. During the night, the jelly melted thus creating a fine mess and food for little creatures living under the floors in the air conditioning chamber below the system. Bugs galore !!!

The RCA 301 was programmable by either machine code or the newly released programming language called Cobol. (see below). Each single machine instruction was composed of ten characters to express four components of usage :

A single character operation code
A single character operation code modifier
A four character source memory address
A four character destination memory address

So a machine instruction like + 4 B144 4038 would mean : ‘add’ ‘four’ numbers (addition from right to left) from ‘memory location 11,144′ into ‘memory location 4038′ and from this you might be able to read the second instruction of this coding sheet (click to grow) :

RCA 301 Coding sheet

Data input and output was either by reading or punching into 7 channel (baudot format) papertape like this: or punch cards with 80 columns like :

80 Column Punch card

though the name Herman Hollerith encoding character sets rings a bell – you can google what part he played in the punch card evolution.

The first compilers for COBOL were subsequently implemented in 1960, and on December 6 and 7, essentially the same COBOL program ran on two different computer makes, an RCA computer and a Remington-Rand Univac computer, demonstrating that compatibility could be achieved. This was my first heavy duty programming language. The cobol ’60 dialect was a core set of language constructs though each hardware vendor added optional elements of syntax to cater for their particular hardware and i/o devices.

In 1997, the Gartner Group reported that 80% of the world’s business ran on COBOL with over 200 billion lines of code in existence and with an estimated 5 billion lines of new code annually. Have calculated that since writing my first lines of cobol in 1962, i have only contributed about 800,000 lines of cobol to this total.

My fourth computer

was the RCA 501, with 40k memory, along with a slave Univac 1005 which was an 80 column card reader or magnetic tape-to-printer system. The 501 was reported to have the fastest sort of it’s day. Due in no small part, to the fact that the tape drives could read in reverse as well as forward. Systems from other manufacturers would need to rewind, or back space then read again while sorting. Since our system has six tape drives ( 7 track 556 BPI ), the sort utility would read data from one drive until memory was full, sort that block and write to each of the five output drives in sequence. A partial merge was then done by reading back a block from each of the 5 drives and doing a partial merge back to the original drive – either over the original data, or more likely, a fresh empty replacement tape mounted before the merge process started. These steps would be repeated until a final merge step would reunite the sorted batches. Perfick !

Univac 1005

Here was a little number cruncher in my back bedroom –

And my second computer ?

Was also a tie for second place between the IBM 7090 transistor systems seen here and it’s slightly earlier twin the vacuum-tubed version IBM704. We used FORTRAN IV as a programming language (formula translation) as the fortran compiler would generate about 10 lines of machine code for each line of fortran syntax – all punched into 80 column punch cards, though i have memories of using telephone-exchange style plug boards to program the 704. Jumper wires were made between two connections to program them. Cobol became available too so it quickly became the developer’s language-of-choice, particularly for business problems.

And my first computer ?

The IBM 1620 – this was the engineering version of the much vaunted IBM 1401 / 1440 business versions. They looked like this:

and the IBM 1620 looked like this:

Fortran 1401 was the language du jour. Allowing a cute 63 pass compiler to transform those statements into P-code, not machine code, as forerunner to virtual machine instruction sets like Pascal that came along some 20 years later. I was lucky in that Cobol became available on the IBM 7090 in late 1966, so that became my primary language of choice – fortran was too hard if you needed to do character manipulations ! Regrettably i never had the chance to learn Algol as the US war in Viet Nam was ramping up, so i was drafted to do that instead. Think i would have rather learned Algol :-)