# What is the "1620's multiplication operation"?

I was stumbling through Wikipedia when I came across the entry for FLOPS, specifically the table in this section.

The first entry is for a computer from 1961, the comment on the right reads

The 1620's multiplication operation takes 17.7 ms.

What is this operation?

I assume it means it can do multiplication in 17.7ms?

From the previous column in that table, the "1620" is the IBM 1620.

https://en.wikipedia.org/wiki/IBM_1620

And yes it means multiplying two numbers.

• Ok so its not some special operation/problem, its just saying it can do multiplication just the same as the IBM 1620 but at 17.7ms? Jul 10, 2014 at 18:51
• It is saying that the biggest "bang for your buck" in 1961 got you a computer that could do a multiplication in 17.7ms. Jul 10, 2014 at 18:55

These stats are for the Model I, and is for two 10 digit numbers. The Model I did in fact use table look ups for math, except for divide (add table was at locations 300-399, multiply at 100-299). The Model II had circuitry for addition and subtraction, but still used table look up for multiply.

As odd as all of that sounds what you would find most odd was that the machine was not fixed word length. That meant that you could add a 100 digit number to a 200 digit number. The other oddity was that each addressable memory location was actually 7 bits, one of them being a parity bit.

The 1620 was my first computer and those two oddities were initially hard to overcome as I moved forward in my career.

The wikipedia section you posted has a reference which links to a document.

That document states that the IBM 1620 Data Processing System could multiply 10 digit numbers in 17,700 microseconds (17,7 ms)

Nor clear whether it's the multiplication of two ten-digit numbers or that the number of digits of both operands add up to ten ( like 1000 * 100000 ).

The IBM 1620 was an inexpensive (lower end model around \$88k USD circa 1959) scientific computer that did all arithmetic (including addition and multiplication) via table look-up. Thus multiplication, requiring multiple look-ups, was quite slow (many milliseconds per). However, computers that could do arithmetic operations faster, at that time, were far more expensive.