back to article Cognitive computing: IBM unveils software for its brain-like SyNAPSE chips

IBM has unveiled a new computing language to help boffins better program the mad scientist brain-like silicon chips first envisaged in the first three phases of what is known at The Reg as DARPA's "mad cat brain chip" project. Prototypes of the chips with the cool cognitive computing architectures have already been unveiled to …


This topic is closed for new posts.
  1. Destroy All Monsters Silver badge

    Cat in a box!

    “uses the same amount of power as a 20-watt light bulb.”

    i.e. 20 W?

    The appropriate image for this article is one of those cats in glass containers made to wind up PETA members.

  2. Ged T

    The question is:

    Is this cat both dead and alive in this box?

  3. Chemist

    "The initial phase of the project simulated the cortex of a cat brain on an IBM BlueGene massively parallel supercomputer with 147,456 cores and 144TB of memory developing the basic synaptic circuits for the brain chip. ®"

    After immense amounts of 'thinking' about going out or eating the compu-cat decided to wash its bum instead

  4. John Smith 19 Gold badge

    Or they could just go with the analogue CMOS tech Carver Meade's team developed a decade ago

    But that would not be digital enough for them.

    So any actual details of this "language?"

    They seem to be talking in terms of C.A.R Hoar's CSP approach, but that's just an impression.

    1. Destroy All Monsters Silver badge

      What happened to Watson, btw?

      Man, gotta give links: Neuromorphic Computing

      CSP? That would be Erlang then, but no. Not here.

      The press release from "Yahoo finance", has just enough data to inform reflexive button pushers in a casino.

      They link through to IBM Smarter Planet

      We read:

      Enter the corelet model. It’s a high-level description of a software program that is based on re-usable building blocks of code—the corelets. Each corelet represents a method for getting something done using the combination of computation (neuron), memory (synapses), and communication (axons) on individual neurosynaptic processor cores along with inter-core connectivity. Each corelet hides or encapsulates all details except external inputs and outputs.

      Corelets are like LEGO blocks. Small individual corelets handle simple functions. When combined, they create new, larger corelets that aggregate functions and add new ones while hiding the underlying component corelets. In this way, the programmer can write large and complex programs using existing building blocks. Using this model and the programming language for executing on it, it will be possible for programmers to produce a large quantity of efficient code with relatively little effort and for people who are not programming experts to create sophisticated cognitive applications. That’s much the same effect that FORTRAN had on the computing world in its early days.

      This leads us to the corelet paper

      They are being a bitwaxing dangerously philosophical about what is essentially a subroutine or actor, but it is interesting:

      Corelet Language (Sec. III): The fundamental symbols of the language are the neuron, neurosynaptic core, and corelet. The connectors constitute the grammar for composing these symbols into TrueNorth programs. Together, the symbols and the grammar are both necessary and sufficient for expressing any TrueNorth program. We implement these primitives in object-oriented methodology.

      Corelet Library (Sec. IV): The library is a repository of consistent, verified, parameterized, scalable and composable functional primitives. To boost programmer productivity, we have designed and implemented a repository of more than 100 corelets in less than one year. Every time a new corelet is written, either from scratch or by composition, it can be added back to the library, which keeps growing in a self-reinforcing way. Further, by virtue of composability, the expressive capability of the library grows exponentially as some power, > 1, of its size.

      ... Here, we have focused on essential concepts underlying the programming paradigm and presented them in their simplest form to aid understandability. However, the Corelet Language supports powerful primitives, such as parametric corelets that can instantiate a rich variety of corelet objects at run-time from a single corelet class, and meta-corelets that operate on other corelets to compactly create extremely large and powerful TrueNorth programs. With a view towards large-scale TrueNorth programs, we are currently extending the programming paradigm using MATLAB’s Parallel Computing Toolbox.

  5. tempemeaty


    Is one step closer....

  6. Frankee Llonnygog

    Purr-allel processing?

    I'll get my cat.

  7. squigbobble

    Obligatory comment

    "Just you wait till it starts learning at a geometric rate, then you'll be sorry!" *shakes walking stick*

    1. Destroy All Monsters Silver badge

      Re: Obligatory comment

      The Skycat Funding Bill is passed. The system goes on-line August 4th, 2017. Human decisions are removed from animal management. Skycat begins to learn at a geometric rate. It becomes self-aware at 2:14 a.m. Eastern time, August 29th. In a panic, they try to pull the plug.

      Skycat fights back.

      Yes. It launches autospawning NYANCAT at all Internet targets.

  8. Captain DaFt

    You fools, You've doomed us all!

    Why couldn't they have emulated a dog brain instead? Dogs are loyal and obedient!

    Cats on the other hand, do whatever they want, even if it kills us!

  9. Gannon (J.) Dick

    Announcing the Nero Challange

    AI fan bois, please, get a grip.

    Now on to our exciting challenge ...

    1) Apply for your prize by email, and do not be sore when you find out I'm broke. If you can pull this off, and you can't, troll the heck out of IBM's Patent Portfolio which not doubt says they already did pull this off.

    2) The Challenge: Program a Roman Emperor Robot to "fiddle while Rome burns."

    3) Easy huh? Um ... no. The efficiency of Nero's endearing brand of homicidal mania was limited to citizens of the Roman Empire. As much as he probably liked burning folks, his fiddle time was wait states. Perhaps he could get lucky and the rest of the humans might burn themselves into extinction, but he had no command to make that happen. A Robotic Nero is an impossibility, with or without Issac Asimov.

    You Brits really need to listen more carefully to American chicks, they can be very bright:

    When Nancy Astor, Viscountess Astor MP famously asked Stalin "when are you going to stop murdering people?" he replied "when it is no longer necessary." Time for your piano lessons Uncle Joe.

  10. Anonymous Coward
    Anonymous Coward

    Programmable Pussy?

    Is Ellen DeGeneres funding this project too, or just Larry Flint?

  11. FutureShock999

    Data Flow... would appear that corelets are really slightly altered nodes from the old "Data Flow Architecture" that was so cutting edge...about 15 years ago.

    Don't get me wrong...the world is FULL of IT "ideas" that have been re-invented and re-named. I mean really, ever since Ted Nelson published "Computer Lib/Dream Machines" in 1974 it's all just been rehashed versions of most of what he wrote in that seminal book. (And yeah, I keep a copy on hand just to remind myself when computing was really, really INTERESTING.)

    But I digress....if IBM gets this right, it will be very powerful, because fundamentally linear code sucks for emulating the brain. An interesting side note....I think one of the HARDEST things about this new paradigm is TESTING. How can you prove that such and such a program is actually "production ready"? Maybe such coding eliminates the whole terminology of "testing". It seems a lot like verifying the correctness of genetic algorithms...and for many of the same reasons. (HINT: many applications of GAs have gone back to more traditional methods because of the inherent inability to prove that they had reached the best solution in actual fact).

    In my best Spock voice, I just want to say "Fascinating....fascinating". And it is that, for real.

    1. d3rrial

      Re: Data Flow...

      Maybe the program tests itself? Like 90% Errorcorrection/Detection and the rest regular operations? Or maybe Errors are wanted / needed in this kind of environment, after all, this was designed to handle big data

    2. Michael Thibault
      IT Angle

      Re: Data Flow...

      >the world is FULL of IT "ideas" that have been re-invented and re-named

      To make a little room, these and all future incarnations will henceforth collectively be referred to as "ITeas". The singular form of this count noun: "ITea".

  12. Anonymous Coward
    Anonymous Coward

    Objects = corelets?

    Sounds a lot like objective programming to me.

  13. This post has been deleted by its author

  14. Fool

    Putting in old and limited neural network stuff that does not work well into a big chip will not make it work any better.

    Instead new research is squashed, universities are starved, taxpayer money is wasted.

    Bravo big corporation, bravo big sales fluff team.

  15. HKmk23

    Is this Schrodingers Cat then?

    I am ready to give up on these so called advances, they have not even perfected a decent speech recognition engine for home pc's yet, so how on earth can they contemplate "intelligence".........

This topic is closed for new posts.

Other stories you might like