We introduce the notion of universal memcomputing machines (UMMs): a class of brain-inspired general-purpose computing machines based on systems with memory, whereby processing and storing of information occur on the same physical location. We analytically prove that the memory properties of UMMs endow them with universal computing power - they are Turing-complete -, intrinsic parallelism, functional polymorphism, and information overhead, namely their collective states can support exponential data compression directly in memory. We also demonstrate that a UMM has the same computational power as a non-deterministic Turing machine, namely it can solve NP--complete problems in polynomial time. However, by virtue of its information overhead, a UMM needs only an amount of memory cells (memprocessors) that grows polynomially with the problem size. As an example we provide the polynomial-time solution of the subset-sum problem and a simple hardware implementation of the same. The practical realization of these UMMs would represent a paradigm shift from present von Neumann architectures bringing us closer to brain-like neural computation.
Name:
sage2014-06-04 2:14
Future of Computing: Universal Memecomputing Machines
Name:
b.u.o.i.s.2014-06-04 2:15
Future of Computing: Universal Memecomputing App
Name:
Anonymous2014-06-04 2:19
GPUs already do that.
Name:
Anonymous2014-06-04 2:28
>>4 Nope. Read it again. GPUs aren't non-deterministic Turing machines. This is a whole different class of computing. If you disagree, you don't know computer science.
Future of Computing: Universal Memecomputing Machines
We introduce the notion of universal memecomputing machines (UMMs): a class of internet-inspired general-purpose joke machines based on systems with memes, whereby processing and storing of humor occur on the same physical template. We analytically prove that the meme properties of UMMs endow them with universal hilarity power - they are satire-complete -, intrinsic sarcasm, functional story telling, and stereotype overhead, namely their collective memes can support exponential punchline compression directly in bitmaps. We also demonstrate that a UMM has the same joke telling power as a non-deterministic meme machine, namely it can solve two-X-walked-into-a-bar problems in polynomial time. However, by virtue of its entertainment overhead, a UMM needs only an amount of meme-readability (memeprocessors) that grows polynomially with the joke size. As an example we provide the polynomial-time solution of the wiped-my-nose-with-used-toilet-paper problem and a simple rage comic implementation of the same. The practical realization of these UMMs would represent a paradigm shift from present joke telling architectures bringing us closer to copycat humor.
On GPU, each shader unit has it's own ALU and memory, so "processing and storing of information occur on the same physical location."
These architectures basically trade frequency for bandwidth. The problem as usual comes with communication between all these computation units. Mammal brain solves communication problem by supplying each neuron with a huge dendritic tree, but that solution is infeasible to replicate on contemporary hardware, because human brain for example requires about 8500000000000000 (8.5 quadrillion) bytes of such memory (given that each neuron has about 100000 connections).
This leads me to conclusion, that modern GPUs use their transistors the best currently possible way.
>>7 Oh look, it's dendrite tree pseudo-intellectual pseudo-scientific fag. The real future of computing will not be based on any stupid brain or neural shit.
Also, I'm just wondering: have you guys ever craved cock so badly that you found yourself running around outside, howling at the moon for it? Literally ROARING at the top of your lungs, wanting nothing less than a dick's head churning against your glottal stop?
Tell me I'm not alone.
Name:
Alexander Dubček2014-06-04 7:49
Future of Computing: Universal Dubs Checking Machines We introduce the notion of universal Dubs Checking machines (UDCMs): a class of post-number-inspired general-purpose computing machines based on systems with repeated decimal digits, whereby processing and storing of information occur on the same physical location. We analytically prove that the memory properties of UMMs endow them with universal computing power - they are Turing-complete -, intrinsic parallelism, functional polymorphism, and information overhead, namely their collective states can support exponential data compression directly in memory. We also demonstrate that a UMM has the same computational power as a non-deterministic Turing machine, namely it can solve NP--complete problems in polynomial time. However, by virtue of its information overhead, a UMM needs only an amount of memory cells (memprocessors) that grows polynomially with the problem size. As an example we provide the polynomial-time solution of the subset-sum problem and a simple hardware implementation of the same. The practical realization of these UMMs would represent a paradigm shift from present von Neumann architectures bringing us closer to brain-like neural computation.
>>12 I'm afraid the only cure for howling cock craving man is a duck turning against his glottal stop 2 times a day for an indefinite treatment.
Name:
Anonymous2014-06-04 8:29
>>12 In an overpopulated world, fertility is the disease. Kill yourself shithead.
Name:
Anonymous2014-06-05 8:08
>>14 In an overpopulated world, the population becomes the disease that consumes itself.
Name:
Anonymous2014-06-05 13:27
In this materialistic world is there any kind of hope? (Nope) Will Catholics ever have a black Pope? I write rhymes like Riddler List them like Schindler Alot of religious figures on this planet are just swindlers Sight beyond sight, those of twilight, the afterlife, a new world is what I like You don't like Commandments but just do them Aight, but when you talk to God again could you give these questions to him? They can't be too long, cause he's a busy man, so tell me about six, and I'll relay em if I can God, (one) Why's my mighty nation lost? (two) Why'd you let Jesus Christ be put on the cross? (three) Why's my people poor and oh so sore? (aiight) In harlem all I see is funeral homes and liquor stores (four) Why'd you let the Indians get used and abused? (five) Why'd you let Hitler crucify all the jews? (alright) some races will never see the best outta life (true) it's like you made a big ass test outta life I got love for you God, but could you explain to me (six) why you took my legs and J and dished out so much pain to me? Don't stress it, kid. It's life, and not even death can seperate the love
Name:
Anonymous2014-06-05 16:44
>>14 Killing a million fertile people won't help if there are a thousand fertile left. You have to kill all of them, then kill yourself.
Also, the US government performed mass sterilization of hapless Muricans in the 20th century, but AFAIK the program was shut down (although with the fascism thriving nowadays, it might have been restarted).
ohn Holdren, Obama's Science Czar, says: Forced abortions and mass sterilization needed to save the planet
Book he authored in 1977 advocates for extreme totalitarian measures to control the population
Forced abortions. Mass sterilization. A "Planetary Regime" with the power of life and death over American citizens.
The tyrannical fantasies of a madman? Or merely the opinions of the person now in control of science policy in the United States? Or both?
Name:
Anonymous2014-06-05 18:55
This smells of Numenta, but I didn't find a connection.
We also demonstrate that a UMM has the same computational power as a non-deterministic Turing machine, namely it can solve NP--complete problems in polynomial time
This is a bold claim, but I can't contradict it. I think it works because the problem of simulating their UMMs is also NP-complete. That's a guess, but easily proven one way or the other.
I nearly flinched at exponential compression, but it makes sense. It is lossy, but no relevant information needs to be lost. Anything relevant almost certainly has a relation to something preexisting in the system which can be used to compress it without loss.
>>7 What are you trying to say? And if you're going to post numbers like that, please say where you got them from. These numbers are basically always wrong but sometimes useful if the source can be taken into account.
I suspect the number of required connections is log(n)-ish. As network mass increases, communication times increase too. I'm not sure if/how that impacts the problem, maybe one of the papers covers it. I get the feeling we could compute NP in approximately P for small enough terms but at some point the tradeoffs start looking more like the standard deterministic computer trying to compute NP, assuming P≠NP.