Brain-like, massively parallel computer made from organic molecules | Hacker News Reader

Brain-like, massively parallel computer made from organic molecules

extremetech.com

74 pointsmrsebastian19 comments

Why the addition of "organic" in the title? or was it removed from the article? I mean, yeah, they're organic molecules: they have carbon in them[1]. So what?

Very neat results, though I note no mention of re-use of the circuits. Anyone know if this is a one-shot deal, or if it's possible to make a general-purpose CPU out of this technique?

edit: the research paper seems to imply the states are all mutable, which suggests to me that this is a very short step away from re-programmable circuits. Awesome. This could get interesting quickly, if there's an easy way to make the circuits in the first place.

[1]: http://en.wikipedia.org/wiki/Organic_molecule

Every time I read an article like this, it makes me a little depressed that I'm still writing code for traditional computer architectures instead of inventing new ones.

I can remember, what 20, no, 30 years ago, when the Connection Machine was going to be the harbinger of a new class of computers. We were all going to have to learn to deal with massively parallel programming.

And then I never heard of it again until years later this (http://longnow.org/essays/richard-feynman-and-connection-mac...) article about Richard Feynman popped up on the internet.

I still don't know why the CM never went anywhere. Inertia, scaling limitations, grabbed by the spooks and now only used by NSA for reading my email?

A number of companies are running into those parallel computing issues again, as we're starting to see more cores added to new CPUs instead of large increases in clock speed like we saw in the late 90's. Plus there are larger distributed computation systems out there like the Amazon and Google clouds.

Politics --- literally. The CM's market niche was high-end research, and most of the sales there were to labs and projects funded by ARPA (or DARPA; the D comes and goes). This led to accusations from competitors that [D]ARPA was steering the labs to the CM, which got a sympathetic ear in Congress --- and all of a sudden, the CM was completely unacceptable to [D]ARPA.

(Accounts of this vary; as an ex-employee and friend of many who remained during this period, the technical staff certainly believed that they were winning those sales with a better product, but those at, say, Kendall Square Research almost certainly believed the opposite, in good faith.)

A possible explanation: http://thedailywtf.com/Articles/Thinking-Machines.aspx

This may or may not be some seriously cool materials science, but parallelizing a cellular automaton is trivial (as cells change state based on local interactions) and has been done with conventional hardware already, as has the construction of systems that change their own wiring in response to experience.

I know very little about the field and would appreciate if you provided some pointers.

Which field? Materials Science? Parallelizing CAs?

Parallelizing CAs.

A Cellular Automaton is inherently parallelizable because the state of any given cell at time t+1 can be computed from nothing more than the the state of its neighbors at time t, see http://www.cs.ox.ac.uk/geraint.jones/publications/book/Pio2/...

thanks

It is false that "parallelizing a cellular automaton is trivial". I can't easily build my own, special-purpose hardware because Intel, AMD, and other such companies control the process, and that's mostly due to the expense of today's silicon foundries being unaffordable to hobbyists. Even if I use the latest FPGAs, such as the Virtex-7 (#1), I am still limited by the underlying technology choices made by Xilinx, as well as by the high cost and continued availability of these high-end parts. I realize that the OP is about some cool work that is merely at the R&D stage, but the bottom-line message is that this could lead to a technique whereby ordinary mortals, albeit ones with (potentially homemade) STMs, could create their own circuitry. It's a pretty big deal if this work results in me being able to build, with relative ease, a huge CA array, upon which I can then apply, through programming, a wide range of computational topologies.

#1: http://news.ycombinator.com/item?id=3162791

"parallelizing a cellular automaton is trivial" != "build my own, special-purpose hardware"

You can parallelize a cellular automaton trivially with any form of parallel computation, from multiple cores, to a GPU, to an FGPA. Hence my skepticism that their application is particularly revolutionary or brain-relevant, as the article seems to imply.

As for FPGAs you can get them fairly cheaply if you aren't looking for a top of the line beast. [1] Given that you can emulate anything up to and including a Pentium on a sufficiently powerful FPGA, I fail to see how you are "limited by the manufacturers design choices" or why you can't "apply a wide range of computational topologies" on one.

In any event, the breathless tone of the article about "brain-like" computing and this somehow being a parallelism silver bullet are unwarranted.

Brain, n. A massively parallel brain-like computer made from organic molecules.

I wonder what the implications of this are in regard to Wolfram's book on automata. Would it be possible to implement an automata computer in a system made from this molecule?

any know of any other related projects to that have hardware that can be played with/hacked?

GreenArrays [1] has 144 chips on a die evaluation boards. The toy is a bit expensive, though.

[1] http://www.greenarraychips.com/

My bet for a brain simulator is on memristor[1] based neural networks. HP is commercializing[2] memristor storage devices as we speak. We need economies of scale such that memristor computing becomes cheap enough to implement universally. Unfortunately, one off research technologies are far too expensive to produce in the necessary quantities for AI research.

[1] http://www.youtube.com/watch?feature=player_embedded&v=Q...

[2] http://www.hp.com/hpinfo/newsroom/press/2010/100831c.html

Reply on news.ycombinator.com