IBM 650

Overview

The IBM 650 occupies a unique place in computer history as the earliest ancestor of the personal computer. Most computers that came before it were designed to expand the envelope of computing, and were usually bigger, faster, and offered more bells and whistles than their predecessors. Often they were designed for large government agencies with insatiable computing needs. The IBM 650, on the other hand was designed to be affordable and easy to use. And compared to what else

was available in the late 1950's, the IBM 650 was:

IBM sold nearly two thousand units of the IBM 650, and it was the first computer to make a significant profit for its manufacturer.

Several types of memory were used in first generation computers: mercury delay lines, electrostatic tubes, magnetic drums, and magnetic cores. Drums were reliable, but slow, due to the inherent rotational delay. Engineering Research Associates, which was acquired by Remington Rand in December 1951, was a leader in drum memory development. The ERA Atlas I, which was delivered to the US Navy in December 1950, had a drum memory of 16,384 24-bit words. Core memory was superior to drum memory, but was initially very expensive, so that for some time the drum continued to be the memory device for small-scale, lowcost systems.
 

The Remington Rand product line needed such a system. Both the UNIVAC I and the 1103 were large systems, selling for about $1 million. IBM's drum memory 650 computer, announced in 1953, sold for $200,000 to $400,000 and was a great success: more than 1800 were sold or leased. IBM licensed the drum memory technology from Remington Rand. Reportedly, Remington Rand settled on that only a few hundred 650s would sell.

Drum Memory

The IBM 650 used a drum memory organized into signed, ten-digit decimal words. The basic IBM 650 had 2000 words of memory. There were 200 read/write heads with 50 words per set of 5 heads, with a later option which added 60 words of core memory where you could store fast loops. Each word could represent a signed decimal integer or an instruction. An optional floating point unit with an eight-digit signed mantissa and a two-digit exponent biased by 50 was available, but the basic unit supported only integer arithmetic, and many users chose to write their own routines to perform arithmetic with floating point (real) numbers.

"bi-quinary" notation

Each digit was represented in seven bit "bi-quinary" notation: one bit out of 5 represented a value from zero to four; one bit out of two indicated whether or not to add 5 to that value, giving the electronic equivalent of the abacus. The front panel had rows of lights in groups of five to display register contents. For example, the integer 281 would be displayed as indicated below, where a display light of the form 0 is considered OFF, and one of the form * is considered ON.
* 0 0 * * 0
0 5 0 5 0 5
1 6 1 6 1 * 6
2 * 7 2 7 2 7
3 8 3 * 8 3 8
4 9 4 9 4 9

 

The complete displays were 10 digits long, with the sign on the right side, just 2 lights in the same space as a full digit.
 

Competition: The Remington Rand SOLID STATE

The Remington Rand UNIVAC Solid State computer was developed in response to the IBM 650. However, the St. Paul division had already announced its drum memory UNIVAC File Computer in January 1955, and Remington Rand management feared that announcement of the Solid State would hurt sales of the File Computer.

The File 1 model with internal program capability finally came out in August 1958. The File Computer was not a success in the marketplace: fewer than 200 were sold. In the meantime, management had allowed sales of the Solid State in Europe, where it was called the Universal Card Tabulating Machine (UCT). Deliveries there started in 1958. Potential customers in the U.S. heard about the UCT and put pressure on Remington Rand to sell it there. The company relented, and American deliveries began in 1959.

The UNIVAC Solid State Computer was priced at $350,000 or leased for $7000 per month. Since the Solid State was faster and more capable than the IBM 650, sales were brisk during 1959 and 1960. In June 1959, Remington Rand announced that it had written an IBM 650 emulator program to ease conversion. But the market life of the Solid State was cut short by the announcement of the IBM 1401 in October 1959. For the same price, it was faster than the Solid State. The pace of sales slowed. Altogether, about 600 Solid State computers were sold.

Another version of bi-quinary coded decimal, in use on the Solid State Computer, used four bits (plus a parity bit) to represent a digit. The bits stood for five, four, two, and one, as indicated in the table below.
5 4 2 1 value
1 1 0 0 9
1 0 1 1 8
1 0 1 0 7
1 0 0 1 6
1 0 0 0 5
0 1 0 0 4
0 0 1 1 3
0 0 1 0 2
0 0 0 1 1

 

IBM 7070

The IBM 7070 was a "transistorized 650." There was a 650 simulator available for the 7070. The 7070 had a machine word of ten decimal digits, plus a sign that could be positive, negative, or alphabetic. Each digit was represented by five bits, coded so that two out of the five were on and three off; if the machine encountered any digit that didn't have two-out-of-five, it halted immediately.

Instead of the 650's drum memory, the 7070 had 10,000 words of core memory. Alphabetic information was coded as two decimal digits, so a word could hold five characters. Physically, the machine was imposing, filling a large air-conditioned room with six-foot high boxes for the CPU and memory, tape drives, online card reader and card punch.

Reliability of core was a concern when early machines were first introduced, and the elaborate two-out-of-five (bi-quinary) coding was designed to meet those concerns.

IBM 650 Machine Code
An IBM 650 machine code instruction was of the form: xx yyyy zzzz
where xx was the op code, yyyy was the operand address and zzzz the address of the next instruction. Thus each instruction contained a jump, to allow for the possibility of "optimization." If you program a drum machine with the instructions stored sequentially, you have to wait at least one drum revolution to read the next instruction. By calculating the expected execution time for each instruction, you can place the next instruction at the correct rotation angle around the drum so that it will come up under a read head when the current instruction is done. Since instructions could execute in as little as little as 0.3 ms (for an add), versus a drum revolution time of about 4.8 ms, careful optimization could increase execution speed by a factor of 5 or more. This is similar to the way in which disk drives use an "interleave factor" which interleaves logically adjacent sectors to improve read/write performance.

Instruction words consisted of a two-digit function code, a four-digit operand address, and the four-digit address of the next instruction. The address of the next instruction was important in a drum memory environment. Since the drum was constantly rotating, it would move some distance while each instruction was being executed. So, to minimize the delay between instructions, it would be best to have the next instruction positioned on the drum at the place where the read-write head was when execution of the current instruction was completed. As a result, instructions which followed each other in program logic would be scattered around the drum, not physically next to each other. The manuals for machines gave instruction timings, so that programmers could try to reduce rotational delays. This approach was called minimum latency programming. It was complicated by the need to fetch operands, so that the programmer had to keep in mind the locations of data and of the next instruction. To aid 650 programmers, IBM published a memory chart. It had 200 rows and 10 columns, with each cell representing a word of memory. As you wrote your program, you would place each instruction and data word in an optimal location and then mark that memory cell off as used on the chart.

It should be noted that pseudo-code interpreters L1 and L2 were designed for the IBM650 by  Bell Labs  in 1955 and 1956 . These were designed to provide a "higher level" interface for programmers, and produced more compact code than the machine's real instruction code.
 

The IBM 650 Open Shop
What really makes the IBM 650 the ancestor of personal computers is that it popularized a concept known as "open shop programming." Prior to the IBM 650, computers were so expensive that most programmers did not have physical access to them. You handed in a card deck containing your program and data and got it back with your output some time later. Often this took several days. Overnight turnaround times were considered good. With the '650, you were actually given a block of time to work with the machine. You could run your program, see what went wrong, fix it and try again. This put control in the hands of the individual programmer. Because of its cost, the machine was kept busy 24 hours a day. Your time might be scheduled at 3 a.m, and as well you might be forced to share your block of time with another programmer. When your program stopped, or went into a loop (you could tell by the way the lights flickered) you were supposed to record the console lights, get off and let your partner use the machine while you looked for the bug. You would have to list your output cards, it there were any, on a IBM 407 accounting machine. The IBM 650 did not have a printer.



Additional resources

This page was produced by Dr. Art Miller for use in the CS 3711 (Programming Languages) course offered by the Department of Mathematics & Computer Science at Mount Allison University,  Sackville, New Brunswick, Canada. As part of the course, students were required to produce an emulator of the IBM 650.