National museum of computing [0] in Bletchley houses a rebuild of the Colossus computer. The people who volunteer there are great and always happy to explain the history.
PS. Bletchley is around an hour on the train from London Euston. Bear in mind that even though the museum of computing is located in the old Bletchley park complex, it requires a separate ticket from the main Bletchley Park museum.
The rebuild sourced the parts from other contemporary machines, and there's a decent chance some of these components came from the dismantling of the original Colossus, so some of the rebuild is probably also original :)
I think the main innovation was the use of valves (similar to tube logic), while other machines of the day used relay logic.
This thing wasn’t really a computer though, in that it was not turing complete, since there wasn’t any support for branching or programmatic examination of its output.
Instead, the operators would tell it what to run, look at the result, then manually apply a flowchart to decide what operation to take next.
These first turing complete machine that “worked” was probably the Z3 from 1941. It was electromechanical and ran at one hz (vs the colossus which ran at 5000hz starting in 1942).
It’s unclear to me if the z1 or z2 were turing complete or not. The z1 supported floating point(!)
This article gives an overview of more machines, and links to articles about most of them:
> I think the main innovation was the use of valves (similar to tube logic), while other machines of the day used relay logic.
Vacuum tubes and Thermionic Valves are the same thing. "Tubes" is the name that stuck in the states. "Valves" in the UK.
> This thing wasn’t really a computer though, in that it was not turing complete,
Turing completeness isn't a requisit to compute. Tommy Flower's big step was taking the electronic switching technology used in telephone exchanges and applying it the computation. Others said that they would be unreliable with the switching speeds involved and mechanical switches were the only way to do it. Flower's knew the key to reliability was thermal stability and designed his machine.
>
> These first turing complete machine that “worked” was probably the Z3 from 1941.
No, it didn't have a condition branch. Z4 had that, but after ENIAC did.
The Atanasoff–Berry computer was even earlier (working prototype in 1942, but much of the design had already been done by 1940).
This was not a general-purpose computer, but a computer made with vacuum tubes that was dedicated to solving a single kind of problem, the solution of systems of linear equations.
The Atanasoff–Berry computer not only used vacuum tubes for computation, but it also included the first DRAM memory, which was made with discrete capacitors.
The report “Computing Machine for the Solution of Large Systems of Linear Algebraic Equations”, written by John V. Atanasoff in 1940, demonstrates a very good understanding for that time of the principles on which electronic computing circuits and memories should be based.
Moreover, the electronic circuits used in the Atanasoff–Berry computer have been a source of inspiration for the later ENIAC, because one of the members of the ENIAC team has made a visit there, inquiring about details of the computer.
While the Atanasoff–Berry computer is likely to have been the first which has used vacuum tubes for computation, it has been derived from a series of British inventions made during the previous decade, which were used for making electronic counters used in nuclear and particle physics research, for measuring ionizing radiation.
The electronic counters had been the first application of digital circuits made with vacuum or gas tubes, before automatic computers. The well-known high speed of the electronic counters is what has suggested the application of vacuum tubes for computing, both in USA with ABC, then ENIAC, and in UK, with Colossus.
Apparently, someone proved the Z3 was Turing complete about ten years ago. The Wikipedia article I linked claims this. I wonder if it was “accidentally” Turing complete, or if it was turing complete when used as intended.
It’s silent on the z1 and z2’s turing completeness, but I don’t think those ever actually worked, sort of making it a moot point.
(Agreed about “computer” being a fuzzy definition - Surely the mercury pool tabulators the US census used could compute stuff, they predated tubes, and maybe even relays.)
Not uncommon. Collosus was classified for 50 years. All examples were broken up and destroyed along with all the plans by order of Winston Churchill. After it was declassified years later, the few people that knew about it set about rebuilding it from whatever fragments of information they could scrape together through old notes that had survived and interviews with the original team. Flowers (the designer) was still alive at the start of the project, but I don't think he lived to see it working.
Apparently, when Collosus became public there were talks given where the Americans who worked on ENIAC (I think) were stunned to find that the British had solved a lot of the same problems they faced just before them. It was an extremely closely guarded secret.
The British and GCHQ kept the stuff secret for ~30 years. They kept it secret from the public only, the Americans knew, and the place was riddled with Soviet spies.
GCHQ also had invented the RSA algorithm before Rivest et al but kept it quiet.
Ah so room scale computing (and its power requirements) are kosher when it's not AI. But involve AI and out the woodwork comes "think about the carbon"
14 comments
[ 3.5 ms ] story [ 45.8 ms ] threadPS. Bletchley is around an hour on the train from London Euston. Bear in mind that even though the museum of computing is located in the old Bletchley park complex, it requires a separate ticket from the main Bletchley Park museum.
[0] https://www.tnmoc.org/colossus
https://en.m.wikipedia.org/wiki/Colossus_computer
I think the main innovation was the use of valves (similar to tube logic), while other machines of the day used relay logic.
This thing wasn’t really a computer though, in that it was not turing complete, since there wasn’t any support for branching or programmatic examination of its output.
Instead, the operators would tell it what to run, look at the result, then manually apply a flowchart to decide what operation to take next.
These first turing complete machine that “worked” was probably the Z3 from 1941. It was electromechanical and ran at one hz (vs the colossus which ran at 5000hz starting in 1942).
It’s unclear to me if the z1 or z2 were turing complete or not. The z1 supported floating point(!)
This article gives an overview of more machines, and links to articles about most of them:
https://en.m.wikipedia.org/wiki/History_of_computing_hardwar...
Vacuum tubes and Thermionic Valves are the same thing. "Tubes" is the name that stuck in the states. "Valves" in the UK.
> This thing wasn’t really a computer though, in that it was not turing complete,
Turing completeness isn't a requisit to compute. Tommy Flower's big step was taking the electronic switching technology used in telephone exchanges and applying it the computation. Others said that they would be unreliable with the switching speeds involved and mechanical switches were the only way to do it. Flower's knew the key to reliability was thermal stability and designed his machine.
> > These first turing complete machine that “worked” was probably the Z3 from 1941.
No, it didn't have a condition branch. Z4 had that, but after ENIAC did.
https://en.m.wikipedia.org/wiki/Turing_completeness
This was not a general-purpose computer, but a computer made with vacuum tubes that was dedicated to solving a single kind of problem, the solution of systems of linear equations.
The Atanasoff–Berry computer not only used vacuum tubes for computation, but it also included the first DRAM memory, which was made with discrete capacitors.
The report “Computing Machine for the Solution of Large Systems of Linear Algebraic Equations”, written by John V. Atanasoff in 1940, demonstrates a very good understanding for that time of the principles on which electronic computing circuits and memories should be based.
Moreover, the electronic circuits used in the Atanasoff–Berry computer have been a source of inspiration for the later ENIAC, because one of the members of the ENIAC team has made a visit there, inquiring about details of the computer.
While the Atanasoff–Berry computer is likely to have been the first which has used vacuum tubes for computation, it has been derived from a series of British inventions made during the previous decade, which were used for making electronic counters used in nuclear and particle physics research, for measuring ionizing radiation.
The electronic counters had been the first application of digital circuits made with vacuum or gas tubes, before automatic computers. The well-known high speed of the electronic counters is what has suggested the application of vacuum tubes for computing, both in USA with ABC, then ENIAC, and in UK, with Colossus.
It’s silent on the z1 and z2’s turing completeness, but I don’t think those ever actually worked, sort of making it a moot point.
(Agreed about “computer” being a fuzzy definition - Surely the mercury pool tabulators the US census used could compute stuff, they predated tubes, and maybe even relays.)
One of my favorites movies: Colosssus: The Forbin Project
https://m.imdb.com/title/tt0064177/
Apparently, when Collosus became public there were talks given where the Americans who worked on ENIAC (I think) were stunned to find that the British had solved a lot of the same problems they faced just before them. It was an extremely closely guarded secret.
GCHQ also had invented the RSA algorithm before Rivest et al but kept it quiet.