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

Supercomputers

Discussion in 'Computer Hardware' started by amusicsite, 23 Jun 2011.

  1. amusicsite

    amusicsite dn ʎɐʍ sᴉɥ┴ Staff Member

    Location:
    UK
  2. Shaun

    Shaun Über Geek

    Starts with F and ends with ... well, you know the end bit.

    I love the comment from MarcBurrows underneath:

    Has Terminator and the Matrix taught us nothing?!

    :roflmao:
     
  3. amusicsite

    amusicsite dn ʎɐʍ sᴉɥ┴ Staff Member

    Location:
    UK
    "electricity bill of about £6.16m a year" - There is one very happy electricity supplier nearby :wink:
     
  4. amusicsite

    amusicsite dn ʎɐʍ sᴉɥ┴ Staff Member

    Location:
    UK
  5. Shaun

    Shaun Über Geek

    Fantastic reading, especially the part about how a Core i5 / i7 in your little home PC can do up to 200 gigaflops (more than the early multi-million dollar super computers!).
     
  6. amusicsite

    amusicsite dn ʎɐʍ sᴉɥ┴ Staff Member

    Location:
    UK
    It was only about 10 years ago that the top apple computer was doing 1gigaflop. How quickly things move on.
     
  7. amusicsite

    amusicsite dn ʎɐʍ sᴉɥ┴ Staff Member

    Location:
    UK
  8. Shaun

    Shaun Über Geek

    Astounding amount of power, but what amazes me more is that so much power is required to simulate fluid dynamics.

    Is it because of the scale they are working at? Are they computing the finest of possible detail in these models?
     
  9. amusicsite

    amusicsite dn ʎɐʍ sᴉɥ┴ Staff Member

    Location:
    UK
    It's complex stuff. Pour milk into coffee and it will spread around the coffee. How and why this happens is ton (maybe even bigger!) of heat exchanges, bumps, knocks, raising, falling type things. In fact fluid dynamics is often linked with random number generation for using in modelling . Because as far as we can tell it's mainly random. So I guess if you want to prove it's not random and work out the code, there is probably quite a bit to crunch. :eek:
     
    Last edited by a moderator: 2 Apr 2014
  10. Shaun

    Shaun Über Geek

    Hmmm ... so each cup of coffee will have similarities, certain sets of things will happen each time you make one, but not in a set way? And fluid dynamics is an attempt to compute enough random possibilities to be able to see the core pattern of set things that happen most of the time?
     
  11. amusicsite

    amusicsite dn ʎɐʍ sᴉɥ┴ Staff Member

    Location:
    UK
    Something like that I think. But not my area of expertise.
     
  12. Shaun

    Shaun Über Geek

    I was thinking about your example and the variables and I started to get an idea of the complexity:
    • Ambient temperature of the room
    • Temperature of boiled water
    • Temperature of mug
    • Temperature of spoon
    • Temperature of milk
    • Volume of water
    • Volume of milk
    • Volume of coffee granules
    • Average size of coffee granules
    • Rate of pour from kettle to mug
    • Rotational stir speed of spoon
    • Direction of stir (clockwise, anti-clockwise, figure of 8, etc.)
    • Fill level of mug
    • Airflow around the mug
    • Size of mug aperture in top

    There's probably a few I've missed off, but randomising just those variables and computing them into thousands and thousands of sets of data that then have to be analysed to provide useful data sort of gives me the idea that it's not as easy as you might first imagine - even for a simple thing like making a cup of coffee! :eek:
     
  13. amusicsite

    amusicsite dn ʎɐʍ sᴉɥ┴ Staff Member

    Location:
    UK
    :coffee:
     
    Shaun likes this.
  14. amusicsite

    amusicsite dn ʎɐʍ sᴉɥ┴ Staff Member

    Location:
    UK
    Or maybe they were just testing on of Dyson's new inventions...
     
    Shaun and DaisyChain like this.
  15. amusicsite

    amusicsite dn ʎɐʍ sᴉɥ┴ Staff Member

    Location:
    UK
  16. Yellow Fang

    Yellow Fang Veteran Geek

    Location:
    Reading
    Fluid mechanics gets very complicated very quickly. It's to do with those Navier-Stokes equations (iirc) which are not actually entirely solvable. Laminar flow is not too unpredictable, but turbulent flow is. That is largely why weather forecasting is so difficult. One of the first uses for computers was weather forecasting. President Eisenhower gave it the go-ahead, remembering how D-Day was nearly cancelled because of the difficulty in predicting the weather. It is why weather forecasting computers and climate simulation computers are so massive.

    I attended an IET or BCS lecture once in which one speaker talked about one of these massively multi-cored computers. A new computer language, based on C, was being developed to take advantage of the parallel computing possibilities. The other speaker talked about the climate model they were developing at the university and why it took so much computing power.
     
    amusicsite likes this.
  17. amusicsite

    amusicsite dn ʎɐʍ sᴉɥ┴ Staff Member

    Location:
    UK
    The momentum equation governs how particles move as a result of inertial, internal, and external forces:​
    [​IMG]
    Tait equation: It is sometimes written as
    [​IMG]
    or in the integrated form
    [​IMG]
    “near density” and “near pressure” :
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    .... All looks dead simple...
     
    Shaun likes this.
  18. DaisyChain

    DaisyChain Active Geek

  19. Shaun

    Shaun Über Geek

    I wonder if parts of it will be reused or applied to different projects? It seems a shame to dismantle such a powerful machine, but looking at the power usage figures I can see why.

    Maybe projects like SETI or Rosetta@home could benefit.
     
  20. beanz

    beanz Staff Member Staff Member

    Let's face it - all our current favourite bits of tech are going to be museum pieces one day soon. Our grandchildren won't have the foggiest how we could have thought they were so cool! :laugh: