What is the true performance of the 2009 Mac Pro?

Discussion in 'Mac Pro' started by neiltc13, May 4, 2009.

  1. macrumors 68040


    May 27, 2006
    I ordered my Mac Pro last week and it should be with me soon. I'm excited to finally get rid of my iMac and be back to a "real" desktop, but I'm a little confused as to how well it performs because there is so much information out there.

    I'm getting the 2.66GHz Quad model, which apparently has the Xeon W3520 processor. I'm also having 3GB RAM and a Radeon 4870.

    The most surprising thing to me were the Geekbench scores. I know this pretty much ignores GPU performance and the numbers that are provided in the blog entry state that the machine I'm getting yields a score of 8144.

    A PC system which is running the Core i7 920 only gets 7503. Is the Mac Pro I am getting really more powerful than this system? I thought Core i7 was easily the best desktop processor money can buy?

    As a gamer, I'm also interested in 3D Mark scores but right now I can't find any for the most recent Mac Pro models. If anyone knows of some, please share!
  2. macrumors 6502

    Jan 16, 2008
    near Cambridge
    There are a lot of bench marks on these forums but you'll need to do a little searching (they tend to get lost in long threads). I think Barefeats have done a few gaming benchmarks. For games you'd presumably run the Pro under Windows so the benchmarks should match any 4870 machine - the processor shouldn't be a problem, it will all come down to the graphics card.

    But to answer your point on core i7, the Xeon chip is a core i7 chip with minor changes such as the ability to use EEC memory so bench marks should be very similar.
  3. macrumors 6502

    Nov 17, 2005
    Halifax, Canada
    The Core i7 920 and the Xeon W3520 are the same chip, except that the latter having support for ECC memory.

    The difference is probably due to the speed of the memory (clock speed, number of channels, etc), the overhead the different OSes add, and whatever else the tester was running (ideally, it should be nothing, but might not be).
  4. macrumors 65816

    Genghis Khan

    Jun 3, 2007
    Melbourne, Australia
    geekbench is good for measuring CPU/RAM performance, but there are slight differentiations for the same models (e.g. my pro gets 9000 consistantly on geekbench).

    also, while the i7 920 runs at the same clockspeed as your mac pro to be, macs tend to be better hardware optimised. (hardware isn't as simple as pluggin in and turning on). to further that note, your mac pro is an i7 server processor (i.e. better).

    as for the 4870 performance, should be the same as on PC, but barefeats should have some good stuff (like this http://www.barefeats.com/nehal05.html , but you cansearch for more if you want)

    also, just as a finishing note, if you're relatively unknowledgeable about mac pro performance and upgrades, barefeats is a fantastic resource.

    congrats on the pro ;)
    - michael
  5. macrumors 601


    Jan 9, 2008
    Slight? He he... No, Massive differences! My mac varies 1000 points or something from the 2nd to the highest ever recorded for this system type to the lower mid-range. And that's on the same machine. ;) Personally I don't think Geekbench is good for anything other than as a kind of a meaningless on-line results competition.
  6. macrumors Penryn


    Mar 23, 2005
    Cuidad de México
    I'd like to see the results the OP gets.
  7. thread starter macrumors 68040


    May 27, 2006
    I found some benchmarks of the Core i7 920 on the 3D Mark site.
  8. macrumors 601


    Jan 9, 2008
    Yeah too bad there's nothing even kinda like that for the Mac (that I know of). A properly written and assembled robust benchmarking suite would be most welcome for OS X. The closest thing we have is Apple's own "Instruments" and that's really better suited for profiling than benchmarking. Until something good becomes available (if it ever does) we're stuck with a bunch of little tests that reveal only very specific elements of our systems - or like GeekBench that actually doesn't work and isn't useful.

    So far I have investigated and used:

    GeekBench 32 and 64
    Disk Speed Bench X
    AJA System Test
    OpenGL Extensions Viewer
    Magnus CPUTester
    and various app specific tests like for PS, LightWave3D, Maya, and etc.

Share This Page