Solent seems to have moved
Last I checked, it was a little closer to Portsmouth than Plymouth
The world’s oldest working digital computer was rebooted on Tuesday following a painstaking three-year refurbishment. The slow-but-steady machine will now be used to educate school-age programmers. The Harwell Dekatron – AKA The WITCH Computer - crunched its first calculations 61 years ago and was used to build Britain’s first …
Easy to forget that the UK was once a world pioneer in Nuclear Power AND Computing. Nice to see we have preserved something from that era so future generations can understand what this country used to achieve.
Plus, I bet it smells nice when running, that authentic, warm, ye olde electricale smelle, unlike modern computers.
All that cloth and phenolic wax insulation warming up nicely. Love the, “impressively terrifying”, power supply, been near a few of those in my time. The third pic reminds me to look out those old spools, sadly I have no punched cards left. It's great to hear about such projects coming to fruition, 'sourcing spares' must have been in an interesting journey. Those were the days, by 'eck lad, youngens will just giggle at grandpa's funny old machines.
"Easy to forget that the UK was once a world pioneer in Nuclear Power AND Computing".
Not to mention aircraft.
By the way, YAHAC, your handle has a typo.
"PUBLISHED: 12:05, 19 November 2012 | UPDATED: 07:58, 20 November 2012"
"The Harwell Dekatron weighs 2.5 tons, contains 10,000 moving parts and can work without a break for 80 hours a week.....
Micheal Bay directs another fast and hardhitting action sequel*
After the destruction of the planet Franchiseatron by the Deathapple(tm) during the Corporate Patent Wars, a lone escape pod crash lands into the far side of Mars (now sponsored by Yagooglehoosambucks) activating machines far older than time below the ** unexplored surface.
Rising from the dust were Machines which cried out for revenge over the fall of something something....
*contents may be less fast or hardhitting than advertised
**current explorations still pending
I for one bow down to our new Overlord - may his reign last for another 64 years
A simple division would take the Dekatron more than 10 seconds to complete, but it was 1,500 times faster than employing a mathematician with a calculator, according to museum director David Hartley.
Sounds as if they needed to get some better mathematicians. Four hours and ten minutes to do a simple division with a calculator?
"Sounds as if they needed to get some better mathematicians. Four hours and ten minutes to do a simple division with a calculator?"
Nah: its the teabreaks, sleeping, eating, and the replication that these mathematicians insist on doing that slows them down. Plus the mistakes reading numbers and the odd fat fingering.
Now, what was the instruction set like? What functions were built in?
PS: I miss the old science museum in Newhall St.
And not in the sense of "plug boards" it's an actual *instruction* set.
Quantum computer developers please take note
BTW I count *three* uses for the Dekatron.
Memory, display and logic (the ability to non-destructively read it and pass it to *another* device)
and if you can *freeze* the system you can do visual debugging.
I'm guessing the speed limit was set by relays, *not* the dekatrons.
Got to wonder if you could "overclock" this?
Used to programme this machine back in the 1960's.
The relays are used to decode instructions and set up the required data paths, basic addition and subtraction are purely electronic limited in speed by the switching speed of the cold cathode decatrons. The electronic parts of the addition process took about 30-40 mS + extra for carry cycles. Multiplication used "shift-and-add" - a bit like long multiplication - and required relay intervention to set up the shifted data paths, the operation was so slow you could actually see the partial products build up. Division was similar although the need to count the number of operations made it even slower. The (shifted) divisor was repeatedly subtracted from the dividend until a sign change was detected, it was then added back once and new cycle of subtractions started with an increased shift - again it was possible to watch the quotient being built up.
Incidentally division by zero was a particular problem as the repeated subtraction of zero never resulted in a sign change. If this happened a "watch dog" circuit noted that no new instructions had been read in for some time and, depending on the setting of a control switch, either blew a hooter or shut the machine down.
Addition and subtraction took place directly between memory locations, multiplication and division proceeded via a special memory location called the accumulator.
The dekatrons are visual and it is possible to stand in front of the machine and determine the contents of any memory location by the simple expedient of looking at it. An excellent feature for the introductory programming courses held at Wolverhampton.
Programmes were normally read from paper tape, each instruction being read from tape before execution, this operation contributed significantly to the slowness of the machine. If you wanted a programme loop you need to make the tape into a physical loop using a paste-pot. There was also a technique for incorporating something called a "block number" in amongst the orders on the tape, this functioned in exactly the same way as a label in a more modern programming language.The original "Gloy" paste pot I used in the 1960's was still in the spares box when the machine moved to Bletchley Park!
It was also possible to store instructions in memory (like a modern computer) but with only 90 memory locations this was rarely done. It was actually slower to read instructions from memory than it was to read them from paper tape. [Not quite sure why].
There were further problems with loops, the mechanical readers are rather heavy handed and after several passes had a tendency to poke extra holes in the tape. Solutions included the use of, expensive, linen based paper tape and repeating the "loop body" multiple times on a larger physical tape. I remember one large loop that was routed round an Anglepoise lamp on the other side of the room.
The most interesting task I programmed on this machine was a key (as in door key) design programme for a local firm of lock and key manufacturers [Messrs Chubb]. The problem was that on a mortice key (the sort with sticking pieces of metal called wards) only certain combinations of ward length were valid, a single ward sticking up above its neighbours would be weak, adjacent wards of the same height were a security problem etc.. Along with another schoolboy (I was let off school for a day a week to use the WITCH) we programmed the WITCH to print out all the valid combinations.
Like all machines of its period, the WITCH had what can only be called "presence", a wall of flashing lights, clacking relays, clattering tape readers and the slightly asthmatic whirr of the output printer springing into life.
Remembering the state the machine was in when it reached Bletchley Park, the fact that it has been restored without any significant replacements is a tribute to both the original designers and the restoration team.
Attending the re-boot ceremony I was given a lapel name badge that described me as "Wolverhampton WITCH", I'm keeping it for Halloween!
Interesting that it could run machine code out of RAM; that would make it one of the very first von Neumann machines in the world.
I wonder what you could achieve with this using modern dynamic-code-generation tricks... probably nothing, but it would be interesting to try. Has anyone done an emulator?
ordinary "orange" (Neon) dekatrons up to a few kilohertz, designed for display. Argon ones are faster. You can buy "NOS" Hydrogen based Dekatrons from Russia, They are very much faster. Some early desktop calculators used Thyratrons and Dekatrons and no relays.
Before Computers there were Computors, people who did quite complex calculations on hand-driven desk-top calculating machines for the purposes of statistical or numerical analysis. I had to clarify the distinction during the entrance interview for my first graduate employment opportunity before answering my new boss' question "Do you know what a computuh is?"
Grief, I remember working at Bearley Radio Station (GPO in those days) using a decatron frequency counter. It was like watching witchcraft.
Then I moved to another company, where I remember feeding paper tape into a ttl-based computer so it could be used for gas-chromatography.
Sigh. Yoof of today.
Back in the 1950s and 1960s Dan Dare lived and worked in London, which was implicitly supposed to be the world capital of technology. Earth's main spaceport was located at Croydon, if I remember aright - although corrections are welcome (and probably inevitable). There was an American called Hank and a Frenchman called Pierre, but never did we see any distinctively American or French technology. It was simply assumed that the RAF would keep getting bigger and bigger and flying further and further...
At last, it's time to confess - I was the person who modified the PDP8 RIM loader tape so that when you tried to load a binary tape, the ASR33 printed "HELP LET ME OUT" and went chattering away dumping the entire (magnetic) core) memory onto paper and ringing the "End-Of-Line" bell.
Ah! - Happy days. Was it 1967 or 1968?
Biting the hand that feeds IT © 1998–2020