Loading ...
Sorry, an error occurred while loading the content.
 

Re: Seeking Language for Parallel Programming

Expand Messages
  • dan michaels
    ... TMC, 64K parallel processors strong, is long road kill. http://en.wikipedia.org/wiki/Thinking_Machines
    Message 1 of 47 , Mar 6, 2009
      --- In SeattleRobotics@yahoogroups.com, Pete Bergstrom <bergstro@...> wrote:
      >
      > Matthew Tedder wrote:
      > > The Propeller? I will look into that. The truth is, multi-threading isn't parallel processing. Multiple cores is, but only once for every core. Such
      > > techniques are not appropriate for fluid dynamics because multiple
      > > simultaneous processes will be out of sync. In any case, large scale parallel processing is far too slow on the fastest possible microprocessors.
      >
      > And that's why Cray is still in business...
      >


      TMC, 64K parallel processors strong, is long road kill.

      http://en.wikipedia.org/wiki/Thinking_Machines
    • David Buckley
      Thank you for posting the HHGG link. I had presumed everyone would know why 42 was a cool number. There seems to be no reason why we couldn t make robot
      Message 47 of 47 , Mar 11, 2009
        Thank you for posting the HHGG link. I had presumed everyone would know why 42 was a cool number.

        There seems to be no reason why we couldn't make robot colonies but the fact is we can't in any real sense of the word. Brooks' subsumption doesn't work and I think only Connell really understood it. Actually Brooks is on record in a letter to Owen Holland at the GreyWalter Archives admitting that subsumption was invented by Grey Walter.

        > I noticed how all of them have essentially gotten to pretty much the same point of ["intellectual"] development and stalled. Pity

        My robots too! Time and day to day living get in the way. My best biped is now over 8 years old.

        David


        ----- Original Message -----
        From: dan michaels
        To: SeattleRobotics@yahoogroups.com
        Sent: Tuesday, March 10, 2009 9:39 PM
        Subject: [SeattleRobotics] Re: Seeking Language for Parallel Programming


        --- In SeattleRobotics@yahoogroups.com, "David Buckley" <david@...> wrote:
        >
        > Actually when I first got an email address david@robots42,free...... I wanted david@... but it was taken, My post code (zip code) was OL4 2PN, and 42 being a cool number a friend suggested I try that, hence the robots42. At the present count I am some way beyond 42 robots.
        >

        Oh ....

        http://www.google.com/custom?q=hitchhikers+guide+42

        > Thank you for Edelman, have to look him up. I don't think that studying the cellular organisation of ant brains is going to tell us a lot about the organisation of more complex brains but it does go to dispel the huge monolithic neural net approach and awaken us to the rich symphony of thought that is being played in and between all the specialist parts.
        >
        > There seems to be no reason why we couldn't make small robots, put them in a closed environment, wireless them to a large network of PCs, and get them to forage for food, to build nests, to take prisoners and use the prisoners as slaves to grow food, to clean and repair nests, to repel invaders etc. We can build the mechanics, we have got tiny cameras thanks to phones, etc. The stumbling block is that nobody has a clue how to make it all work.
        >

        Yes, we certainly should build small robots using ant brains for inspiration. Kind of like the past 20 years of subsumption robots, in fact. But then we have to move to the next-step eventually. The problem is to find it.

        >
        > At the weekend I was listening to the radio where some experts were describing the symbolism behind the songs of Leonard Cohen and their relevence to life and I wondered if Cohen understood what they were >talking about especially when he was writing the songs.
        >

        Cohen's songs and poetry [I liked them a lot in my youth] mostly had a very Freudian/Oedipusean undertone. I really doubt there was much else. LOL.

        >
        > We can describe what we think is going on and describe the outcomes but until we can build build robots which work like the robots we imagine then in reality we haven't got a clue.
        >
        > Personally I think it is just an engineering problem but it is a robotics engineering problem, not mechanical engineering, not electrical engineering, not electronic engineering, not computer engineering, not software engineering...... but robotics engineering and most people (especially young graduates) just haven't got the breadth of knowledge neccessary and they don't know they don't know.
        >

        The foibles of youth. One has to be burned a few times in order to learn.

        >
        > One thing is for sure, we didn't get from not being able to fly to putting men on the Moon by trying just one design of aeroplane or relying on simulations. We need to build robots, lots of them, which are more complex accumulations of stuff we know works. Then we can try out more advanced thinking - algoithms etc. Going back to music, I don't think anybody wrote a symphony without first learning to play scales.
        >

        Your posts actually caused me to count how many bots I have built - esp after I wrongly guessed what 42 meant. LOL. I counted 14 so far, but on review, I noticed how all of them have essentially gotten to pretty much the same point of ["intellectual"] development and stalled. Pity.

        >
        > There has been some posting on vision, now I for one have not a little trouble trying to understand a scene by only using one stationary eye and I guess the same goes for most people, so how come most everybody (until recently) tries to do machine vision that way?
        >

        Good realization. However, it will help a great deal to bob your mono-vision head back and forth, and up and down, like birds are very prone to do. Many items in the scene will suddenly take on new definition.

        > David - robots42
        >
        >
        >
        >
        > ----- Original Message -----
        > From: dan michaels
        > To: SeattleRobotics@yahoogroups.com
        > Sent: Monday, March 09, 2009 10:09 PM
        > Subject: [SeattleRobotics] Re: Seeking Language for Parallel Programming
        >
        >
        > --- In SeattleRobotics@yahoogroups.com, "David L Buckley" <david@> wrote:
        > >
        > > What you have just described as thinking is really some deductions from (I presume) some measurements on some brains and is really akin to descriptions of how some electric signals seem to go along some metal links between some strange interleaving of metal and silicon in these things called chips. Trying to deduce the program running in a computer from limited descriptions of electrical activity in the circuitry is virtually an impossible task even for the simplest microprocessor. Applying the same process to brains is just going to be even more futile especially when on a day to day task to task basis it is quite obvious that most of what humans do isn't done on any conscious level.
        > > Creating huge arrays of interconnections and hoping it will think, isn't going to work.
        > >
        >
        > Yes.
        >
        > >
        > > The problem is that humans are particularly bad at understanding how they think because most of the time they are manipulating languagised versions of their perceptions rather than acting on and controlling their perceptions.
        > > What we need to understand are the algorithms of the 'thinking' which takes place in ants and flies brains and this is not the same as descriptins of the electrochemical activity. How is they do what they do? If we understood that then we could by degrees build more complex thinking systems.
        > >
        > > David - robots42
        > >
        > >
        >
        > Hey David - robots42 [42 refers to how many bots you have previously designed and built, I should hope?].
        >
        > Years ago, when I was studying neuroscience in grad school, I used to think the same thing you say above, as regards studying the brains of "lower" organisms, and I dabbled with frogs a lot back then.
        >
        > However, anymore, I'm not so sure, and this gets back to your statement above that I responded "Yes" to. A lot of people think that enough complexity in parallel will magically produce intelligence. Very doubtful. However, to follow up, you might be interested in checking out any of the many books written by Gerald Edelman on brain operation. Especially note the material on anatomy / organization.
        >
        >
        >
        >
        >
        > [Non-text portions of this message have been removed]
        >





        [Non-text portions of this message have been removed]
      Your message has been successfully submitted and would be delivered to recipients shortly.