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

GPU Technology..

Discussion in 'Hardware' started by asje1, Dec 10, 2008.

  1. asje1

    asje1 Byte Poster


    Over the last couple of weeks, at work all people have been talking about is Nvidia's new Tesla cards...

    So I was invited to the European launch of Nvidia Tesla Cards and the new 'Personal Super Computer'...

    As a trainee (and a very noobie one at that).. I was a little scared and put off because lets face it I knew everyone there was going to be at least 15 years older than me, be professors or people much higher up the pay scale to me. - And i wasnt wrong!

    But I did enjoy the experience. Now to the perpose of this topic. I sat through some lectures and I learnt that, with the new Tesla cards there is the opportunity for Universities, Hospitals, Labs etc to be able to have access to a 'personal super computer'. These personal super computers are like nothing I have seen, ranging from around 5k-10k, and are as big as a normal sized workstation. Where as some of you may know, Cray used to have really old, BIG, expensive solutions... which isnt practical.

    So the lectures continued, it soon came clear that by using the GPU along side the CPU to do number crunching (in nooby terms) you can speed up whatever your software process is by a staggering x100+... all in a standard sized work station, affordable, same power consumption as a normal workstation etc...

    For example: alot of scientific companies are intrested in this technology because they use GPU for DNA processing or folding etc...

    1 Case study explained how they took CT scans and turned them into 3D models... so 2d bone scans into computer rotatable models... normally this process took weeks to proform a model reconstruction, but by using the GPU (Telsa cards) which is measured in Gigaflops, Teraflops or Petaflops... This would take a day or hours... Which is quite amazing considering what can now be acheived. Professors will be able to do 100's more tests per week, where as before they may have based results on only 100 tests due to time, now they can have much more accurate results, quicker and cheaper!!

    So to allow the Telsa cards to work along side the CPU there is a compiler tool called CUDA, which is an extension the the C language... this allows the hardware to comminicate and speed things up...

    For those who may be intrested here is the Link on nvidia's website Link

    Does anyone else have any intrest in GPU // CUDA etc, im not an expert but i think this new technology is amazing, so i thought we could maybe talk about it as im sure this will be a big hit for the future.

    Certifications: A+, N+
  2. dmarsh

    dmarsh Terabyte Poster

    Not used it but looked at the CUDA stuff on their site about 6 months back, interesting that they even have a Black–Scholes example.


    Increased Parallelism is only going to get more important unless a few fundamental computer architecture laws get broken.

    Amougst others Intels also been working on various solutions, one being better library support.


    So yes concurrency is going to be a big thing...
  3. wagnerk
    Highly Decorated Member Award

    wagnerk aka kitkatninja Moderator

    I'm kinda into my graphic cards and the GPU on them, and from a selfish point of view I want to know how my games graphics (present & future) will look like :)

    Certifications: CITP, PGDip, BSc, HNC, LCGI, PTLLS, MCT, MCITP, MCTS, MCSE, MCSA:M, MCSA, MCDST, MCP, MTA, MCAS, MOS (Master), A+, N+, S+, ACA, VCA, etc... & 2nd Degree Black Belt
    WIP: MSc in Tech Management
  4. dmarsh

    dmarsh Terabyte Poster

  5. asje1

    asje1 Byte Poster

    Lol, like us all. The new tesla doesnt even have a display outlet so i doubt its going to be good for your gaming :P...

    Only going to get better though i guess.
    Certifications: A+, N+

Share This Page