It costs a *lot* to make this version. As we've said right up front, we're working on a lower cost plastic version. (We released this version because a number of folks wrote to us and said that they'd rather have the CNC-cut wooden version even though they knew it would cost a lot more.)
Sir, may I offer my congratulations at the manner in which you came up with the contour logic which process the multiplication function. Beautiful contour design. I love the manner in which the three distributors binary bits create the four paths to service the four digits in the number stored in the memory register , and those holes, to reject the ball if the digit is Zero. The manner in which the contours around the number stored in the multiplier quotient are routed such that the "rejected balls" would count the quotient where each rejected ball triggers the distributor is fascination to behold. I made counters and adding machines, but never anything as "brainy" this. Again Congratulations.
It's a computer. No need to put that in quotes. The computer you might use at home is no different, just the switches are much much much smaller and it uses electrons instead of marbles.
Well, I'm still four years late, but ya know, I vaguely remember having a plastic version of this too. Way back in the sixties. I remember having quite a bit of fun setting it up for a calculation and getting the right answer... Memories, they're a good thing! Thank you! I guess you can tell, I'm getting old. You're only as old a you let yourself be though, so I'm not too old!
Placed my order. The DigiComp II is a significant teaching tool and won't change with time. The interest of those in this forum already proves this point. Makes most sense to have plastic model for the masses, but for the more serious crowd, this machined wood version, and even an aluminum plate version will have an audience. The built quality and engineering is top notch, so the working parts offer additional level of detail as conversation piece. Patiently awaiting my unit...
The multiplication algorithm isn't very efficient and doesn't scale well, because the bigger the number, the longer it takes to multiply by it (the computer needs to do that many additions, so, for example, multiplying by 7 requires 7 additions). A better approach is to "shift & add": for each bit set in the multiplier, you shift the multiplicand by this bit's position number and add up to the accumulator. This requires no more multiplications than the number of bits of the multiplier register (worst case; usually it takes much less, because it adds only for the bits which are actually set). This makes me wonder if it could be possible to modify this machine to use this algorithm for multiplication instead. But it would require a *shift register* for the multiplicand, and some way to control it by the bits of the multiplier register. Any ideas about how such a shift register could be implemented? Another problem with this computer is that it adds the numbers one bit at a time, due to the clock mechanism, which can release only one ball at a time. A better approach would be to use more balls released at once, which could go in parallel through these bits at one machine cycle. Still another problem is that the machine cannot "program itself": the user needs to input the numbers and select the mode of its operation by hand every time before running the computer. The result of one computation cannot be automatically loaded to the input registers to be able to be used in another (subsequent) computation. That's because the state of the flip-flops cannot be "read", and there is no "feedback loop" between the output and the input. But there is one advantage of this machine over other marble calculators I've seen so far: it doesn't use the marbles for _storing_ information, only for _transmitting_ it and _driving_ the mechanisms. This way is better, because if a marble is used for storing information, it can be used only once (single-read memory), and afterwards the information is lost, because the marble needs to go out from the register when it is being read. One cannot "copy" information that way, to make a multiple-read memory. In this approach, on the other hand, the register can be read as many times as we please, as long as the switch stays at the same position (blocking, or allowing the marbles to go through it). I wonder, though, how could we merge the two mechanisms: the flip-flops and the "program" switches (which cannot move on themselves) so that they could be set/cleared by the marbles themselves and then to be read out multiple times by another marbles without changing their state. The change of state should be done only when we want it, on "write" / "update" signal, but not on the "read" signal. Any ideas?
Miah Elf Yeah, sounds so simple if you already have one made by someone else... I wonder what would you do when on a deserted island with no computers and electricity around ;) and the only thing you could use for calculations were coconuts and palm wood. Would you be smart enough to build a full-blown arithmetic-logic unit out of coconuts? ;J
Miah Elf Yeah, because ending up on a deserted island is a barrel of laugh... Why one would need math on a deserted island? Well... for counting days and calculating what year it is, or to calculate your position according to the positions of stars to find out where the hell the island is, or to make some calculations for another devices you try to build, e.g. a radio transceiver to send an S.O.S. signal and get the hell outta there? Or you can just sit there and wank all day long to the rest of your life... It's your life anyway...
My dad first gave me "Dr. Nim" (a version of the "Nim" game, based on the Digi-Comp series) back in the mid sixties, then blew me away a couple years later by giving me Digi-Comp II. I don't know exactly what was on his mind, except that he was jazzed by cool kit; and it would make a fun DIY project (it was a build-it-yourself kit with what seemed like a gagillion pieces, as I recall... I would love have one again - maybe it's time to get the plans...
The base goal is to get below $112, the inflation-adjusted cost of the original 1960's model. That's realistic for a sturdy, well-made plastic version and still about $600 less than originals sell for, these days. If we made it as flimsy as the original, it could cost as little as half of that; we have many tradeoffs to consider in terms of quality versus cost. As for the time estimate, pick any two: Fast, Good, Cheap.
are there any predictions pertaining to the cost of the new version? I remember reading on the website that this plastic version will be released in late 2013. I don't know much about the manufacturing business but why does it that that long?
Sir, I congratulate you on your demonically clever machine. Many moons ago, we half considered a similar machine in which logic gates were water valves controlled by hydraulic pressure, but the cost of the valves put the idea out of contention. Rent and mortgages, you know the score.
We don't have videos of the other operations, but you can read the user guide. Addition can be done at the same time as multiplication: The operation is essentially this: Accumulator = Accumulator + (Memory Register X Multiplier Register). If you set the value of the Multiplier Register to 1, then it performs simple addition.
+Ethan Holter Sometimes this speed of calculation is enough, if it could calculate for you when you can work on something else in the meantime and when you don't need to know the result immediately. Otherwise, you would have to perform these calculations yourself, wasting your time. (Yeah, for small numbers it might seem pointless, but expand it to 64 bits or more and the benefits start to grow ;) ).
+Ethan Holter Also this machine isn't the best one we could come up with. This particular multiplication algorithm is not very time-efficient, because the bigger the number you want to multiply, the longer it takes to do it (the computer needs to add the memory register to accumulator that many times). This could be done faster if we could "shift & add" the memory register instead: for each bit set in the multiplier, shift the memory register left to the position of that bit and then add the memory register to the accumulator. This would require a maximum of B additions, where B is the number of bits of the multiplier register. (But this is a worst-case: normally it would take less additions, because it adds only for each bit actually set in the multiplier.) Currently I'm thinking on how to improve this machine to be able to do that.
people need to watch things like this to understand how things like the Antikethera(spelling bad) mechanism was made. when you know nothing about it it looks like a silly toy, when you see that its set up and you have to learn how to read it a special way to use it you see how painstaking calculatons went into the creation of the device, and it has specific limits.
My dad owns the real, original one. We bought it at a flea market for just 10$. No idea how much it is worth since there doesn't seem to be much of a market for them. Still gotta fix up a few pieces of it though, doesn't fully work. Any idea?
I do want this to get done pretty fast, but I do have to say, I want it to be 75% Good, 20% Cheap, and 5% Fast. Because good quality is better than anything.
Bon bon, I am working on a marble computer that will do the shift and add method. My computer uses a whole other method of setting the bits to on or off.
+Karvák Ferenc Increasing the tilt angle will make the balls roll faster. Also, you can multithread it by pulling the start lever more than once-- although not all operations are thread safe.
Just giving an idea but, do you think you could "sell" the instructions so people could make this themselves. You guys would still get some money from it. It's just an idea.
Imagine disliking someone's hard work because you didn't like the music being played during the demonstration, gee, a person who did that would be a true douchebag
If I were to purchase one of these, does it play that annoying music while it's in operation?
+Bobbie Bees Sorry, no option for built-in music. :)
+Evil Mad Scientist would be funny to see a futuristic pinball themed one though
+Arthur Kimm Let's see it!
+Bobbie Bees That's the Evil part of evil mad scientist I guess.
+Arthur Kimm Are you 8 years old and made something more amazing than this? (Your upload videos) Where's the link?
So it's a wooden binary calculator?
It costs a *lot* to make this version. As we've said right up front, we're working on a lower cost plastic version.
(We released this version because a number of folks wrote to us and said that they'd rather have the CNC-cut wooden version even though they knew it would cost a lot more.)
Sir, may I offer my congratulations at the manner in which you came up with the contour logic which process the multiplication function. Beautiful contour design.
I love the manner in which the three distributors binary bits create the four paths to service the four digits in the number stored in the memory register , and those holes, to reject the ball if the digit is Zero.
The manner in which the contours around the number stored in the multiplier quotient are routed such that the "rejected balls" would count the quotient where each rejected ball triggers the distributor is fascination to behold.
I made counters and adding machines, but never anything as "brainy" this. Again Congratulations.
Brilliant system. Really well done.
That's really cool. It would be a fun device to teach students about binary data!
You're telling me I can have a WOODEN PC some day?? Damn
TheSphongleface Not only that, but it's run by marbles!
Haha, are you saying it would use marbles for pixels if I wanted a screen?
If you mean that clear ones that have colored marbles behind them that roll when color needs to be changed, then Hell yeah!
Can this "COMPUTER" become self-aware and destroy humanity?
I can't rule it out with certainty, but the video is about five years old and it hasn't done so thus far.
It's a computer. No need to put that in quotes. The computer you might use at home is no different, just the switches are much much much smaller and it uses electrons instead of marbles.
Maybe if it looses its marbels
At $349, this is one of the most expensive calculators that I have ever seen.
But at least it's lightning fast...
Well, I'm still four years late, but ya know, I vaguely remember having a plastic version of this too. Way back in the sixties. I remember having quite a bit of fun setting it up for a calculation and getting the right answer... Memories, they're a good thing! Thank you! I guess you can tell, I'm getting old. You're only as old a you let yourself be though, so I'm not too old!
Placed my order. The DigiComp II is a significant teaching tool and won't change with time. The interest of those in this forum already proves this point. Makes most sense to have plastic model for the masses, but for the more serious crowd, this machined wood version, and even an aluminum plate version will have an audience. The built quality and engineering is top notch, so the working parts offer additional level of detail as conversation piece. Patiently awaiting my unit...
Now use floating levers connected to the binary output to add the sum and plot it on a ruler.
The multiplication algorithm isn't very efficient and doesn't scale well, because the bigger the number, the longer it takes to multiply by it (the computer needs to do that many additions, so, for example, multiplying by 7 requires 7 additions). A better approach is to "shift & add": for each bit set in the multiplier, you shift the multiplicand by this bit's position number and add up to the accumulator. This requires no more multiplications than the number of bits of the multiplier register (worst case; usually it takes much less, because it adds only for the bits which are actually set). This makes me wonder if it could be possible to modify this machine to use this algorithm for multiplication instead. But it would require a *shift register* for the multiplicand, and some way to control it by the bits of the multiplier register. Any ideas about how such a shift register could be implemented?
Another problem with this computer is that it adds the numbers one bit at a time, due to the clock mechanism, which can release only one ball at a time. A better approach would be to use more balls released at once, which could go in parallel through these bits at one machine cycle.
Still another problem is that the machine cannot "program itself": the user needs to input the numbers and select the mode of its operation by hand every time before running the computer. The result of one computation cannot be automatically loaded to the input registers to be able to be used in another (subsequent) computation. That's because the state of the flip-flops cannot be "read", and there is no "feedback loop" between the output and the input.
But there is one advantage of this machine over other marble calculators I've seen so far: it doesn't use the marbles for _storing_ information, only for _transmitting_ it and _driving_ the mechanisms. This way is better, because if a marble is used for storing information, it can be used only once (single-read memory), and afterwards the information is lost, because the marble needs to go out from the register when it is being read. One cannot "copy" information that way, to make a multiple-read memory. In this approach, on the other hand, the register can be read as many times as we please, as long as the switch stays at the same position (blocking, or allowing the marbles to go through it).
I wonder, though, how could we merge the two mechanisms: the flip-flops and the "program" switches (which cannot move on themselves) so that they could be set/cleared by the marbles themselves and then to be read out multiple times by another marbles without changing their state. The change of state should be done only when we want it, on "write" / "update" signal, but not on the "read" signal. Any ideas?
+Bon Bon Maybe with some kind of computer... lol
Miah Elf Yeah, sounds so simple if you already have one made by someone else... I wonder what would you do when on a deserted island with no computers and electricity around ;) and the only thing you could use for calculations were coconuts and palm wood. Would you be smart enough to build a full-blown arithmetic-logic unit out of coconuts? ;J
Bon Bon Wtf am I doing math on an island for again? hahaha
Miah Elf Yeah, because ending up on a deserted island is a barrel of laugh...
Why one would need math on a deserted island? Well... for counting days and calculating what year it is, or to calculate your position according to the positions of stars to find out where the hell the island is, or to make some calculations for another devices you try to build, e.g. a radio transceiver to send an S.O.S. signal and get the hell outta there?
Or you can just sit there and wank all day long to the rest of your life... It's your life anyway...
I haven’t seen this video in years.
I would’ve loved to be able to try one of these back in its hay day.
My dad first gave me "Dr. Nim" (a version of the "Nim" game, based on the Digi-Comp series) back in the mid sixties, then blew me away a couple years later by giving me Digi-Comp II. I don't know exactly what was on his mind, except that he was jazzed by cool kit; and it would make a fun DIY project (it was a build-it-yourself kit with what seemed like a gagillion pieces, as I recall... I would love have one again - maybe it's time to get the plans...
how do i still find this old stuff?
The base goal is to get below $112, the inflation-adjusted cost of the original 1960's model. That's realistic for a sturdy, well-made plastic version and still about $600 less than originals sell for, these days.
If we made it as flimsy as the original, it could cost as little as half of that; we have many tradeoffs to consider in terms of quality versus cost.
As for the time estimate, pick any two: Fast, Good, Cheap.
I wonder what it would take to go even further and have it run programs from memory
How do I get my brain to work in a way that will allow me to invent stuff like this?
It still solved it faster than I did.
Cool, but does it play solitaire?
Does it blend?
I'm supposed to be revising. How did I get here.
This is a good demonstration of a finite-state machine.
are there any predictions pertaining to the cost of the new version? I remember reading on the website that this plastic version will be released in late 2013. I don't know much about the manufacturing business but why does it that that long?
Sir, I congratulate you on your demonically clever machine.
Many moons ago, we half considered a similar machine in which logic gates were water valves controlled by hydraulic pressure, but the cost of the valves put the idea out of contention. Rent and mortgages, you know the score.
This is a replica of the original digi comp that was from the 80s or 90s
Though it is clever
I want to see the other operations. And I wonder how easy it would be to add addition to this?
We don't have videos of the other operations, but you can read the user guide.
Addition can be done at the same time as multiplication: The operation is essentially this: Accumulator = Accumulator + (Memory Register X Multiplier Register). If you set the value of the Multiplier Register to 1, then it performs simple addition.
A very good, very big IC chip. Shows how to build if-then-else statements, I want one.
At first I thought that the number of balls was an input. Cool.
its the worlds slowest calculator. at least its faster than the ones in minecraft
+Ethan Holter Sometimes this speed of calculation is enough, if it could calculate for you when you can work on something else in the meantime and when you don't need to know the result immediately. Otherwise, you would have to perform these calculations yourself, wasting your time. (Yeah, for small numbers it might seem pointless, but expand it to 64 bits or more and the benefits start to grow ;) ).
+Bon Bon you have a good point
+Ethan Holter Also this machine isn't the best one we could come up with. This particular multiplication algorithm is not very time-efficient, because the bigger the number you want to multiply, the longer it takes to do it (the computer needs to add the memory register to accumulator that many times). This could be done faster if we could "shift & add" the memory register instead: for each bit set in the multiplier, shift the memory register left to the position of that bit and then add the memory register to the accumulator. This would require a maximum of B additions, where B is the number of bits of the multiplier register. (But this is a worst-case: normally it would take less additions, because it adds only for each bit actually set in the multiplier.) Currently I'm thinking on how to improve this machine to be able to do that.
fuck you
LizardOnATrain & Fricious : confused pony face
TOO MANY DRUM MACHINES IN THE SONG GODDAMN
Music* a song has vocals.
people need to watch things like this to understand how things like the Antikethera(spelling bad) mechanism was made. when you know nothing about it it looks like a silly toy, when you see that its set up and you have to learn how to read it a special way to use it you see how painstaking calculatons went into the creation of the device, and it has specific limits.
My dad owns the real, original one. We bought it at a flea market for just 10$. No idea how much it is worth since there doesn't seem to be much of a market for them. Still gotta fix up a few pieces of it though, doesn't fully work. Any idea?
I've seen them go for as much as $700 on eBay, even not fully working.
Interesting, I haven't seen them at all on ebay! We've owned it for several years now.
I do want this to get done pretty fast, but I do have to say, I want it to be 75% Good, 20% Cheap, and 5% Fast. Because good quality is better than anything.
Bon bon, I am working on a marble computer that will do the shift and add method. My computer uses a whole other method of setting the bits to on or off.
how can i OC it? :)
+Karvák Ferenc
Grease the tracks so the balls roll faster?
Yonkage increasing wolrd gravity! or send this calculator to jupiter
+Karvák Ferenc Put it in a centrifuge to increase force on balls.
+Karvák Ferenc Increasing the tilt angle will make the balls roll faster. Also, you can multithread it by pulling the start lever more than once-- although not all operations are thread safe.
just think of it: there are thousands of these inside a computer
Very excellent demonstration!
cool but, can it run crisis???
How is that even possible??
Music?
What's the music?
awesome, great for educational purposes.
Just giving an idea but, do you think you could "sell" the instructions so people could make this themselves. You guys would still get some money from it. It's just an idea.
I want one badly.
ThePiGuy24 Gaming me 2
Can it play crysis?
This reminds me of the old Amazing Dr Nim game.
Not coincidentally, that was also invented by the "late" ESR company which designed the original 1960s Digi-Comp 2.
I would love to have one, are they for sale?
Jeffrey Nelson yes!
Evil Mad Scientist And where can it be bought?
+Jeffrey Nelson Check the site in the description
Can someone explain this? My mind think this was a game.
Mind=Blown.
I want one...
Now if we could just shrink that down to pocket size. Wait? What! I see what you did there😜😋
in which country do you live
That is really cool.
slowest calculator ever
But it's extremely impressive considering it's just a marble track
I sense jealousy
amazing !!!
Can this be used to cheat on tests?
If you can find a way, I'm sure that everyone will be very impressed.
so cool!
This reminds me of that dr. Nim game...
They're closely related: The Digi-Comp II and Dr Nim were both described in the same patent application.
impressive.
i got high before i watched so i am soooo Lost now LOLOLOL
amaaaaazoooooooooooong!!!!
can it mine cryptocurrency?
Gonna guess that it's not the most efficient method for doing so.
music doesn't fit well. maybe techno would be suitable since its computer logic related.
Bring this to a math exam and see who fails
You would still be on question one after every hands their tests in
nice
Digicomp *TWO*...... first edition
This is a replica of a thing called the Digi-Comp II. (There is also a Digi-Comp I.)
Nergasm. That's awesome
Like nummer 1000
ok
B...b...bu...b.b.b.b...b.bbb...BUT....WHYYYYYYYYYYYYYYYYY?
purchase? (yay 100th comment!)
Needs pipelining.
WAY too expensive
wantz
very practicle...
jk really cool
:)
mas ou menos
ear cancer... tumb down
Imagine disliking someone's hard work because you didn't like the music being played during the demonstration, gee, a person who did that would be a true douchebag
1st
cool