bit-tech.net

TNMOC's EDSAC rebuild uncovers unique circuit diagrams

TNMOC's EDSAC rebuild uncovers unique circuit diagrams

EDSAC technical leader Chris Burton with John Loker and Andrew Herbert show off unique circuit diagrams saved from landfill nearly sixty years ago.

The National Museum of Computing (TNMOC) is celebrating today following the discovery of some of the earliest diagrams of a computer, drawn 60 years ago in an effort to document the world's first general-purpose computer EDSAC.

The National Museum of Computing announced its intentions to host a rebuilt project for the Electronic Delay Storage Automatic Calculator (EDSAC) system back in 2011, and in 2013 began to produce the first parts. EDSAC has a special place in computing history: built immediately after World War II by a Cambridge-based team led by Sir Maurice Wilkes, EDSAC was the world's first practical general-purpose digital computer. It also served as the blueprint for the Lyons Electronic Office I (LEO I), the first business computer and forerunner to today's highly-computerised world.

Sadly, EDSAC itself is no longer with us. The system's more than 3,000 thermionic valves - having pre-dated the transistor - spread over 12 racks were dismantled and spread to the wind in 1958 after nine years of service. Its successor, EDSAC 2, was an improvement in almost all respects, but the historical impact of EDSAC is not to be underestimated - and it's this impact that the Museum wishes to highlight with its rebuild project, to occupy a whopping 20 square metres of the facility when it is finished in late 2015.

To aid the team in its efforts, former University of Cambridge Mathematical Laboratory engineer John Loker has donated 19 detailed circuit diagrams detailing various parts of EDSAC after saving them from landfill. 'I started work as an engineer in the Maths Lab in 1959 just after EDSAC had been decommissioned. In a corridor there was a lot of stuff piled up ready to be thrown away, but amongst it I spotted a roll of circuit diagrams for EDSAC,' explained Loker of his discovery. 'I'm a collector, so I couldn't resist the urge to rescue them. It wasn't until I visited TNMOC recently and learned about the EDSAC Project that I remembered I had the diagrams at home, so I retrieved them and gave them to the [EDSAC Rebuild] Project.'

'Thankfully, the documents confirm that the reconstruction we are building is basically correct, but they are giving us some fascinating insights about how EDSAC was built and show that we are very much in tune with the original engineers: both teams have been exercised by the same concerns,' added Andrew Herbert, leader of the EDSAC Project, of the diagrams. 'Importantly, the drawings clearly show that the aim of EDSAC's designer, Sir Maurice Wilkes, was to produce a working machine quickly rather than to create a more refined machine that would take longer to build. The refinements could come later - and many did as the sequence of diagrams over the five-year period shows.'

Many of the diagrams were created after EDSAC had been constructed in an apparent effort to document the build and aid in the creation of its successor, and show modifications from the original design including circuit improvements that boosted the signal strength and error correction algorithms capable of distinguishing errors in programming from malfunctions of the machine itself. Coverage of 'initial orders,' the equivalent of an initial program loader (IPL) or boot ROM in modern systems, also highlighted that an approach rejected by the rebuild team was in fact the one chosen by the original creators - something the team will now correct.

The rebuild project is due to complete in late 2015 with a fully-working recreation of EDSAC as close to the original as possible, which will be open for public viewing at the Museum. Currently, visitors to the site can enjoy a work-in-progress exhibit also documented in detail on the official website.

5 Comments

Discuss in the forums Reply
Corky42 25th June 2014, 11:01 Quote
Not that i am questioning the article, I'm just hoping to be educated

A TV program i watched claimed America built and run the first GPC, obviously this article calls that into question and further research says it was the ENIAC in 1947.

So have i been misinformed, duped by the TV, or have i misunderstood the label used to describe a general purpose computer ? Were the EDSAC and the ENIAC totally different animals ?
Gareth Halfacree 25th June 2014, 11:12 Quote
Quote:
Originally Posted by Corky42
Not that i am questioning the article, I'm just hoping to be educated A TV program i watched claimed America built and run the first GPC, obviously this article calls that into question and further research says it was the ENIAC in 1947.
It's something historians argue over. The general consensus is that while work was happening near-simultaneously on both projects, ENIAC went online first and is therefore the world's first general-purpose digital computer; EDSAC fans counter with the fact that ENIAC was arguably specialised for its military uses and was impractical to reprogram for other purposes, making EDSAC the world's first practical general-purpose digital computer (as the article states). Naturally, the American's don't like that argument - but the fact that EDSAC would lead to LEO, the very first commercial computer, lends the argument weight.

EDIT: To add more detail, one of the biggest differences in practicality was that EDSAC was a stored-program computer, while ENIAC was not. Imagine if every time you switched your computer on you had to input the operating system and your chosen application manually. Using toggle-switches, or even jumper leads. That's where you were with ENIAC, while EDSAC could simply load a program into memory from punched tape. Here's a Wikipedia explanation:
Quote:
Originally Posted by Wikipedia
ENIAC could be programmed to perform complex sequences of operations, including loops, branches, and subroutines. The task of taking a problem and mapping it onto the machine was complex, and usually took weeks. After the program was figured out on paper, the process of getting the program into ENIAC by manipulating its switches and cables could take days. This was followed by a period of verification and debugging, aided by the ability to execute the program step by step.
Contrast with EDSAC, which was simply "stick paper tape into reader and hit the MAKE PROGRAM GO button." Want to use ENIAC to do something else? Expect a couple of weeks at best to change programs. EDSAC? Pop to lunch and the operators will have it ready and running by the time you get back.
Corky42 25th June 2014, 11:49 Quote
Well i never, thanks for enlightening me ;)

So it seems it's not who was first but more if ENIAC could even be classed as a GPC, it seems to me from your excellent explanation it was not. Well unless you consider having to spend a few weeks getting it to do something else.

To think we used to complain about slow computers twenty odd years ago, just think of having to spend weeks just to change a program. :D
jb0 26th June 2014, 08:30 Quote
Quote:
Originally Posted by Gareth Halfacree

EDIT: To add more detail, one of the biggest differences in practicality was that EDSAC was a stored-program computer, while ENIAC was not. Imagine if every time you switched your computer on you had to input the operating system and your chosen application manually. Using toggle-switches, or even jumper leads. That's where you were with ENIAC, while EDSAC could simply load a program into memory from punched tape. Here's a Wikipedia explanation:
Quote:
Originally Posted by Wikipedia
ENIAC could be programmed to perform complex sequences of operations, including loops, branches, and subroutines. The task of taking a problem and mapping it onto the machine was complex, and usually took weeks. After the program was figured out on paper, the process of getting the program into ENIAC by manipulating its switches and cables could take days. This was followed by a period of verification and debugging, aided by the ability to execute the program step by step.
Contrast with EDSAC, which was simply "stick paper tape into reader and hit the MAKE PROGRAM GO button." Want to use ENIAC to do something else? Expect a couple of weeks at best to change programs. EDSAC? Pop to lunch and the operators will have it ready and running by the time you get back.

That's actually somewhat misleading. The "The task of taking a problem and mapping it onto the machine was complex, and usually took weeks" is program development. the modern equivalent is coding software. This is not going to be appreciably different between ENIAC and EDSAC.

"After the program was figured out on paper, the process of getting the program into ENIAC by manipulating its switches and cables could take days " is actual system configuration time once the program is developed, and this is where EDSAC improved upon ENIAC, both in terms of speed and reliability.

Granted, a few days to "install" a program is still kind of ridiculous for general usage, and it pales before EDSAC's load times(which I expect to have been along the lines of a few minutes, though I seem unable to find actual numbers), but it's much shorter than a couple of weeks.




Personally, I feel that the use of "practical" as a modifier in these sorts of comparisons is USUALLY just a way of wiggling around pre-existing work to claim you're first at something.
But in EDSAC's case, I think it's fair, to the extent that it's unfortunate people wanting to claim they were first at something without actually BEING first at something have poisoned the word practical.




And ENIAC, of course, is commonly misrepresented as the first computer instead of the first general-purpose computer(general-purpose is, of course, purely a representation of functionality, not practicality)
Gareth Halfacree 26th June 2014, 08:57 Quote
Quote:
Originally Posted by jb0
That's actually somewhat misleading. The "The task of taking a problem and mapping it onto the machine was complex, and usually took weeks" is program development. the modern equivalent is coding software. This is not going to be appreciably different between ENIAC and EDSAC.
True enough, that, and something I should have been clearer about - especially with my misleading closing paragraph.
Quote:
Originally Posted by jb0
Granted, a few days to "install" a program is still kind of ridiculous for general usage, and it pales before EDSAC's load times(which I expect to have been along the lines of a few minutes, though I seem unable to find actual numbers), but it's much shorter than a couple of weeks.
Remember, though, that while the paper-based development can take place while the system is running another program, the actual programming requires the system be down for several days.
Quote:
Originally Posted by jb0
Personally, I feel that the use of "practical" as a modifier in these sorts of comparisons is USUALLY just a way of wiggling around pre-existing work to claim you're first at something. But in EDSAC's case, I think it's fair, to the extent that it's unfortunate people wanting to claim they were first at something without actually BEING first at something have poisoned the word practical.
I certainly think it's fair, otherwise I'd never have included it in the article. The direct development from EDSAC to LEO kinda proves that; J. Lyons & Co. would never have tried to use an ENIAC-style system for business purposes, because it was entirely impractical.
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums