Think you really understand the progress in computer performance? Take a look

In 1976 the Cray company introduced the world's first supercomputer.  The Cray-1 weighed 10,500 pounds and had a 115,000-watt power supply.  It was by far the fastest computer in the world.  

It cost over $40 million in 2024 dollars and used $35,000 of electricity per month.

Twenty-two years later (1998) the same company built a supercomputer that could perform one *trillion* "floating-point math operations per second," which techies shorten to "1 teraflop."

Unless you're a techie you have no idea what that means, but in 1998 it was really impressive.

Just four years later (2002) Cray rolled out the follow-on, which was 50 times faster.  So, a 50-fold increase in performance in just four years.  But the cost in today's dollars was still over $40 million.

Today a top-end desktop computer can do 80 to 100 teraflops, and costs maybe $2,000.  So it's about twice the speed of a multi-million-dollar supercomputer of 2002 and costs not just one percent of the price of the supercomputer of 2002, but less than one one-hundredth of one percent of that price.

Most of us are accustomed to leaps of computer technology, but such a huge improvement in the performance-to-cost ratio is stunning.  

That was a year ago. Today a single-board computer just 2.2" wide, weighing just three ounces, using five watts of power--is 4.5 times faster than the Cray 1.  It sells for $100.

Now let's look at storage:  Back in 1978 most personal computers stored data on 5-inch flexible disks, coated with the same material as magnetic recording tape.  These so-called "floppy disks" could typically store 360,000 characters.

Today a chip just over a quarter of an inch on a side can hold 256 BILLION characters--the equivalent of 700,000 floppies.

What does this mean to you?

We're slowly finding out.