Vacuum Valves and Mercury Memory: Remembering Sir Maurice Wilkes

Your rating: None Average: 5 (2 votes)
4,231 Views
Maurice Wilkes inspecting a mercury delay line memory.
Copyright Computer Laboratory, University of Cambridge. Reproduced by permission.

The semiconductor industry is constantly producing new innovations that enrich our lives. But, we stand on the shoulders of giants. Sir Maurice Wilkes, who passed away on November 29th this year, was one of the tallest.

Sir Maurice is credited with several pioneering developments in computing, but is best known for leading the team at Cambridge University in England that developed EDSAC, the first truly modern computer, which ran its first program in 1949.

The UK’s Daily Telegraph newspaper wrote an excellent obituary to Sir Maurice which tells the stories of both Sir Maurice and EDSAC. It’s well worth a read.

EDSAC was not the first computer; that is generally credited to Charles Babbage and his never-finished Difference Engine. It wasn’t even the first electronic computer. That was ENIAC, developed at the University of Pennsylvania and designed to calculate artillery firing tables. EDSAC was, though, the first computer that resembles what we use today – one that stores both program and data in temporary storage, allowing any program to run without reconfiguring the hardware. Before EDSAC, the computer that Sir Maurice’s team built, reprogramming a computer took days and meant flipping thousands of switches and moving cables around. EDSAC could instead load a program from paper tape, making the machine the first modern computer.

It’s interesting to compare EDSAC with, say, a modern laptop computer:

  EDSAC
Laptop
Logic elements
3,000 vacuum valves, 1” diameter 600,000,000 transistors, 32nm gate length (100 billion would fit in the same area as one valve)
Operations per second 650 Approx. 3 billion
Footprint 12 racks, one small room One human lap
Memory type 32 Mercury delay lines, 576 bits per tube Solid state DRAM, 1 bit per cell
Memory capacity 18,432 bits 34,359,738,368 bits (4 gigabytes)
Input 5-hole punch tape electronic keyboard
 Output teleprinter 15” LED-backlit liquid crystal display

Source: Daily Telegraph and Wikipedia.

It would be a mistake to dismiss EDSAC as a fossil. The level of ingenuity is staggering. Today, we solve problems with brute computing force. Throw enough processors and memory at a problem and you can solve anything. It’s an impressive approach, but not an elegant one. Back then, every bit counted. With less memory capacity than you’d find in a cheap digital watch today, EDSAC found a 79-digit prime number and solved differential equations in gene theory.

The memory technology is simply astonishing. A mercury delay line was a long tube filled with liquid mercury. Data was stored as acoustic waves rushing back and forth in the liquid. Hundreds or even thousands of bits could be stored at any one time. Today, we store information in the nanoscale capacitors of a DRAM chip, one bit a time.

An unexpected (and unfortunate) side effect of this breakthrough was the discovery of debugging. In his memoirs, Sir Maurice said: “I can remember the exact instant when I realised that a large part of my life from then on was going to be spent in finding mistakes in my own programs.”

Today’s nanoengineering is amazing, of course, but it’s far too small to see. Mercury sloshing back and forth in a tube and the orange glow of racks of vacuum tubes? Now that’s impressive.

Bookmark/search this post with:

Comments

EDSAC

This article is nicely balanced between today's technical wizardry and the pioneering work of Sir Maurice Wilkes. It is only right that we should marvel at the achievements of the recent past, and not dismiss them as primitive.

Post new comment

The content of this field is kept private and will not be shown publicly.
Be sure to start the URL with "http://" or "https://" as appropriate.
CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.
Image CAPTCHA
Copy the characters (respecting upper/lower case) from the image.