Speed difference between the 2.8Ghz 4core and the 3.33Ghz 6 core?

Discussion in 'Mac Pro' started by iondot, Aug 27, 2010.

  1. iondot macrumors member

    Joined:
    Jun 30, 2008
    #1
    Does anyone know what the actual real world speed between the 2.8Ghz 4core and the 3.33Ghz 6 core?

    I'm specifically curious how this plays out with Photoshop and Final Cut, but I'm suspect other people may be curious about lots of programs.
     
  2. iondot thread starter macrumors member

    Joined:
    Jun 30, 2008
    #3
  3. dissolve macrumors 6502a

    Joined:
    Aug 23, 2009
    #4
    Keep an eye on Barefeats. They currently have compared the quad to 6-core with Portal and a few CPU benchmarks. In the past, they have used used Photoshop and other software to evaluate differences as well.

    The 6-core is going to be faster at basically everything. A more 'fair' comparison may be between the 3.2 quad and the 6-core. At least then you'll only see substantial differences in applications that are aware of multiple cores.

    EDIT: links are helpful

    http://barefeats.com/wst10.html
    http://barefeats.com/wst10g.html
     
  4. VirtualRain macrumors 603

    VirtualRain

    Joined:
    Aug 1, 2008
    Location:
    Vancouver, BC
    #5
    For CS5, here's a graph from MPG that you can use to interpolate the performance of the 2.8GHz Quad...

    [​IMG]

    The 3.33GHz Hex scores 20.2 seconds
    The 3.33GHz Quad scores 22.4 seconds
    The 2.8GHz Quad should score 26.6 seconds (16% lower clock speed)

    For FCP, which is not multi-threaded without the use of compressor, the 2.8GHz Quad will perform 16% slower than the 3.33GHz Hex in most tasks. That means importing an AVCHD movie and transcoding to Apple ProRes that takes 5 minutes on the 3.33Hex, will take about 6 minutes on the 2.8GHz Quad.

    The interesting thing is that the $1200 uplift to purchase the Hex is probably not worth it unless your time is precious. A more difficult decision is whether the $400 uplift for the 3.2GHz Quad is worth it. :)
     
  5. eponym macrumors 6502

    Joined:
    Jul 2, 2010
    #6
    Barefeats does indeed.

    [​IMG]

    [​IMG]
     
  6. VirtualRain macrumors 603

    VirtualRain

    Joined:
    Aug 1, 2008
    Location:
    Vancouver, BC
    #7
    True, but neither of those is in any way helpful in determining performance with CS5 and FCP.
     
  7. englishman macrumors 6502a

    englishman

    Joined:
    Nov 6, 2006
    #8
  8. MT0227 macrumors 6502

    Joined:
    Jun 16, 2009
    #9
    Right now, for most users, clock speed still trumps the number of cores you have as most programs hardly take advantage of more than one.

    Per Mac Performance: "A 3.33GHz model runs 13.6% faster than 2.93GHz, and 38.7% faster than 2.4GHz. A 13.6% clock speed difference means about 68 seconds vs 60 seconds, or 1.13 seconds vs 1 second — noticeable, but not decisive. But 3.33GHz vs 2.4GHz is substantial eg 83 seconds vs 60 seconds, or about 1.4 seconds vs 1 second — now that you'll notice."

    Given this math, the 2.8GHz would come in at 71 seconds vs 60 seconds, or 1.18 seconds per 1 second. The first question you need to determine is will you notice this.

    If you plan to run multiple CPU intensive processes that will take advantage of additional cores the above theme changes. Depending on the number of processes you run together on average would determine how many cores you should be looking to purchase.

    One thing to keep in mind, the Quad-Core processors take 1066MHz DDR3 ECC SDRAM , the 6-Core 1333MHz DDR3 ECC SDRAM. Most seem to think the 3.33 6-Core is the best overall bang for the $$. It gives you the added clock speed on single threaded applications, while providing the added cores for multiprocessing and applications that are/will (future) taking advantage of multi core processors.

    Good luck
     
  9. eponym macrumors 6502

    Joined:
    Jul 2, 2010
    #10
    I agree for most people clock speed will trump in general. But clock speed and cores aren't the only factor, even in poorly optimized apps. Memory capacity and bandwidth is very important if you need it (and those that do usually know it ;))

    Take the MPG huge memory intensive CS5 test as an example: the slow 8-core trounces the 6-core by almost a minute. And I would imagine the 12-core would absolutely destroy all of them.

    @ OP

    I'd wait and see regarding FCP—don't try and factor its performance on the machines right now. As I understand it, the app is due for an update and is currently not doing much to use the horsepower available to it. But if you look at apps like Logic and After Effects, you'll see the difference optimization makes. Logic on the 8-core outperforms the 6 and AE is neck and neck. Photozoom scores on MPG are also another example of the difference optimization makes.

    It would be reasonable to assume that next version of FCP will be way faster on the 6-core than the 2.8 quad. The 6 is faster now and will be even faster as time goes on. If you can afford it, it's a better machine.
     
  10. iondot thread starter macrumors member

    Joined:
    Jun 30, 2008
    #11
    I understand your logic, but find it frustrating that I have to play guessing games with the FCP development team. Adobe has been very disappointing in this regard. CS5 shows little improvement over CS4 in making uses of dual processors and multiple cores.
     

Share This Page