mac pro for a research project

Discussion in 'Mac Pro' started by marvinGS, Feb 2, 2009.

  1. marvinGS macrumors newbie

    Apr 10, 2008
    Hi all,

    I am considering to buy a mac pro to the university I am working for (well, actually the university is buying, I am just deciding what to buy) to be used in a research project. It will mainly be used to compile C and for MATLAB. Considering the upcomming snow leopard and the grand central technology, I want to get a good GPU too. My question is: Is a quadro 5600 fx worth the extra $2460 over 4x HD 2600 XT regarding computational power?

    Thanks in advance.
  2. Tallest Skil macrumors P6

    Tallest Skil

    Aug 13, 2006
    1 Geostationary Tower Plaza
    Since you can't combine the power of multiple cards, the four HD 2600 XT are worthless unless you are using seven to eight monitors.

    We don't know much of anything about Snow Leopard, so we can't tell you if having four crap cards will be better than having one amazing card. Sorry.
  3. alanlindsay macrumors member

    Mar 4, 2008
    I would recommend going with the standard video card for now and consider upgrade benefits later after SL is released. I'm a math grad student who uses Matlab, C and F77 frequently and I really can't see the GPU being tapped any time soon. Unless your work is so computationally intensive that it is beneficially to write it specifically for the GPU from the get go (e.g. protien folding etc).

    I think you should know also that there is a serious serious problem with Matlab and Leopard. Specifically, I'm using Matlab 2008b and have a MBP (early 2008 2.4) with Ubuntu 8.10, XP, 10.5.6 triple booted and I get 3x the performance in Linux that I do in Leopard ( from bench ). There is something very wrong with Matlab in Leopard...others have noticed it too but I'm not aware of a solution. Something you may want to consider.

    In conclusion my advice is the following. Do not upgrade the GPU for now but wait until the upcoming technologies have demonstrated that they can effectively utilize the GPU in your line of work - then consider upgrading. If you are doing serious computations in Matlab - load it in Linux or XP to enjoy a significant speed boost. You can also use the Intel C/Fortran Compiler in Linux free for non commercial use. This compiler is significantly better than gcc.


Share This Page