1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Hardware Editing Memory and Multi-core Programming

Discussion in 'Article Discussion' started by Tim S, 8 Nov 2009.

  1. Emon

    Emon New Member

    Joined:
    14 Jun 2004
    Posts:
    680
    Likes Received:
    0
    Uh yeah, software engineers learn all that stuff, too. In practice, it is not easy to keep track of for anything other than trivial applications. Try writing a massive, distributed server application and handle all the concurrency with only locks and semaphores, see how far that gets you.
     
  2. kgwagner

    kgwagner New Member

    Joined:
    20 Jan 2005
    Posts:
    7
    Likes Received:
    0
    Having spent a great many years programming, I'm always amused by these people who want to "educate" programmers, or "do more research" into how to program when they themselves aren't programmers.

    It's certainly difficult to do parallel programming. But, that's not why it isn't done more. Hard-core programmers dearly love a challenge, particularly if it involves some unusual problem solving. That's not the issue, though. The real issue is that many things simply can't be done in parallel. You need the results of the previous operation in order to proceed with the next operation. There's simply no way to start on the next operation otherwise. The vast bulk of problem resolution is a serial process. If you need to find E and it depends on finding what D-C is, and you don't know what C is until you subtract B from A, then you're screwed. E is going to have to wait until the first couple operations are done, and they'll have to be done one after another. I don't care how many processors you have, if you don't know what A-B is, you won't know C, and until you know that, you can't subtract C from D, so you can't find E. Looks like three operations, and it is, but you can't save any time trying to use three processors to complete those operations. In fact, it'll probably COST you time.

    But, let the wishful thinkers throw money at the problem. The programmers will make a lot of money trying to show them the obvious until it finally dawns on them.
     
  3. crazyceo

    crazyceo New Member

    Joined:
    24 Apr 2009
    Posts:
    563
    Likes Received:
    8
    "WHOOOOOOSSHHHHHHH!!!!"

    That was the sound of this all going over my head!

    This is all far too technical for me but if I could just say that the very fact some very intelligent people seem to be "working" very hard to unlock the full potential of multicore cpu's, whilst playing Quake. I think they deserve all the kudos in the world for getting Microsoft to pay for that.

    And guys, who the hell paints a building pink?
     
  4. Landy_Ed

    Landy_Ed Combat Novice

    Joined:
    6 May 2009
    Posts:
    1,428
    Likes Received:
    39

    +1

    Most points raised in the thread are valid, but horses for courses, we may as well be arguing that blue is more useful than red or yellow. Any hardware level assistance that gets us there quicker is welcome in my book, even if it's often just a case of every service using it's own core (and of course we have shared memory bandwidth to worry about, but that's another challenge.....). Just as hardware advances, programming techniques need to as well. We're looking forward, here, not back.

    @crazyceo, maybe they have an odd crush on miss hoolie from balamory! It's not as bad as a pink Range Rover (there's one of those driving around my county, & that's just SICK...big man-car in girly pink, it's just not right)
     
  5. crazyceo

    crazyceo New Member

    Joined:
    24 Apr 2009
    Posts:
    563
    Likes Received:
    8
    Yeah didn't that slapper Jordan want one of those? or was that a Pink Bentley?
     
Tags: Add Tags

Share This Page