iMac faster than Mac Pro? Seriously???

Discussion in 'Mac Pro' started by macenforcer, Aug 15, 2008.

  1. macenforcer macrumors 65816

    macenforcer

    Joined:
    Jun 9, 2004
    Location:
    Colorado
    #1
    Got a mac pro 2.8ghz 8 core with 16gb ram and 8800GT card. Ran xbench against my new 2.8ghz 24" imac. Imac scored 178.45, Mac pro scored 177.88. WHAT?

    So I ran an encoding test. The mac pro finished 30% faster but still, thats not that much faster for having 8 cores. What gives? Is the mac pro just hype???
     
  2. Umbongo macrumors 601

    Umbongo

    Joined:
    Sep 14, 2006
    Location:
    England
    #2
    They use the same processor technology. The differences are the FSB and the prefetch routines (which I doubt show up in xbench), two processors also run slower than 1 at same speeds. It's all about bandwidth and you wouldn't notice the difference. One benchmark run isn't conductive to the iMac always being faster, I'd expect to see similar scores for each system though. Run benchmarks that take advantage of 16GB of memory and the advantages of the 8800GT and you can easily see the MP being better.

    You don't have to go far on this forum to know that more cores are only beneficial on a limited number of applications. If you are running Applications that can fully utilize 4+ cores or your workflow can really use that 16GB of memory then it isn't an issue.

    I'm not really sure what hype you are refering too.
     
  3. macenforcer thread starter macrumors 65816

    macenforcer

    Joined:
    Jun 9, 2004
    Location:
    Colorado
    #3
    That the Mac Pro is leaps and bounds faster than an imac hence the double price tag.

    So I used quicktime to do the encoding test. Is quicktime not multithreaded? Seems to be using all 8 cores but only 25% of each one and it fluctuates. The imac uses 95% of both cores the whole time.

    So I suppose the only difference is

    imac 1066mhz bus, 2 cores, 4gb 800mhz memory
    Mac Pro 1600mhz bus, 8 cores, 16gb 800mhz memory

    Both have the same hard drives in them and ofcourse the graphics cards..


    Mac pro is still a great machine. I was just hoping it would wipe the floor with my imac but apparently not. So there you go folks. iMac is just fine for 90% of the stuff you need to do. I wish there was a multithreaded benchmark app out there that would take the cores into account, but I really fear its the bandwith of the system bus that is the real enemy here. The only speed advantage is that the mac pro has 2 buses instead of 1 in the imac. The next mac pro which eliminates the bus should be dramatically faster I would expect.
     
  4. Artful Dodger macrumors 68020

    Artful Dodger

    Joined:
    May 28, 2004
    Location:
    In a false sense of reality...My Mind!
    #4
    If you have something to render from Blender or Maya try doing it on both and you will see a big difference not to mention you can choose how many cores you want/need/have using Blender (not sure about Maya).
    Run the same size file on both the iMac then the Mac Pro using equal cores. After a few test runs, max out the Mac Pro, no hype needed, real world times will be much better on the Mac Pro :D As many have stated before it all depends upon the App. as to how it can handle your workflow for lack of a better term.
    On the bright side, 20-30% faster is still faster and you will get more done in less time which hopefully means more $$ to you and others ;)
     
  5. Darth.Titan macrumors 68030

    Darth.Titan

    Joined:
    Oct 31, 2007
    Location:
    Austin, TX
    #5
    A benchmark is a benchmark. I never have put much stock in them. Real-world use is a much better indicator.
    As to the Quicktime encodes, from what I've read Quicktime is not multithreaded very efficiently and from experience I can tell you it is the slowest encoder I've ever used.
    Try your encode test using something like VisualHub or HandBrake and I bet you'll see a more pronounced difference between the iMac and Mac Pro.
     
  6. Infrared macrumors 68000

    Infrared

    Joined:
    Mar 28, 2007
    #6
    Try GeekBench :)

    If it's only 30% faster, that suggests to me the test is poorly
    multi-threaded.

    I have run encoding tests that show a massive improvement
    with 8 cores compared to 4 or 2 cores. I'm pretty sure a MP
    would crush an iMac running those tests.
     
  7. goinskiing macrumors 6502a

    goinskiing

    Joined:
    Jun 25, 2008
    Location:
    Meridian, ID
    #7
    Wait until Grand Central is out and then you'll be very happy with the 8 cores. :cool:
     
  8. Infrared macrumors 68000

    Infrared

    Joined:
    Mar 28, 2007
    #8
    Well, it's a ravenous monster if fed right. But that
    "if" is critical. You wouldn't feed a shark lettuce,
    now, would you? o_O

    That suggests that QT is suboptimal for what you've
    done. You might try switching between codecs and see
    if you get better performance.

    For example, I generally use FFmpeg for encoding. Its
    performance usually far exceeds what Quicktime Pro is
    capable of. However, I did notice that it's slower with
    '-vcodec mpeg4' when using 8 threads than it is using 1.
    Looking in Activity Monitor I see that the cores aren't
    really being hammered. It's much like what you observed:
    only 25%-ish of each being used.

    Put simply: the threading is rubbish with that codec.
    However, if I use '-vcodec libx264' 8 threads gives an
    impressive performance boost. Which is nice, because
    H.264 encoding is where you really need a speed boost.
     
  9. rockinrocker macrumors 65816

    rockinrocker

    Joined:
    Aug 21, 2006
    #9
    yep, it all depends on the software. if it can't use all those juicy cores, then it's not going to do much better then a dual.
     
  10. FF_productions macrumors 68030

    FF_productions

    Joined:
    Apr 16, 2005
    Location:
    Mt. Prospect, Illinois
    #10
    Since some programs aren't fully optimized to stream along 8-cores, that's why you run multiple instances/encode multiple things at the same time...
     
  11. xraydoc macrumors 604

    xraydoc

    Joined:
    Oct 9, 2005
    Location:
    192.168.1.1
    #11
    Handbrake is very well coded for multiple cores. Pegs all 4 in my machine at 95-100%; crunches a 2 hr Video_TS folder in to H.264 in about 35 minutes.
     
  12. Firefly2002 macrumors 65816

    Joined:
    Jan 9, 2008
    #12
    XBench isn't a good benchmark for any system, multiple cores or not. Its results are highly varied, and shouldn't have any stock put in them.
     

Share This Page