A Peak at E5 Mac Pro performance

Discussion in 'Mac Pro' started by xgman, Apr 6, 2012.

  1. xgman macrumors 601

    xgman

    Joined:
    Aug 6, 2007
    #1
    From an IGM article today:
    http://www.tonymacx86.com/viewtopic.php?p=348208#p348208

    Hacintosh benchmarks of Xeon E5 CPUs foretells next Mac Pro speeds?
    April 6th 2012

    A forum post over at tonyx86.com has a Hackintosh system running the new Xeon E5 processors. The CPUs are expected to be in the next update to the Mac Pro. The specific CPUs used are dual Xeon E5 2690, which includes 8 cores each at 2.9 GHz. So, a 16 core machine.

    Geekbench scores came in at 41,000 running in 64-bit mode under Windows 7. Unfortunately the forum poster hasn't gotten the OS X side running to full capacity yet. Likely, things will work better once Apple pushes out an OS update to accompany new Mac Pros.

    To give some perspective, the current top end Mac Pro is officially recorded at 24,159. The current top end iMac is listed at 12,575. Such a Mac Pro upgrade suggests some impressive CPU crunching. But, hold on to your wallet, the E5 2690 retails for $2040 each. That's around $600 more than the original MSRP of the X5670 in the current top of the line Mac Pro. Even with volume discounts, Apple will tack on their margin, which all suggests the new Xeon E5 processors may push Mac Pro prices even higher.
     
  2. goMac macrumors 603

    Joined:
    Apr 15, 2004
    #2
    Yet more proof the iMac is just as fast as the Mac Pro.

    /s
     
  3. Tutor macrumors 65816

    Tutor

    Joined:
    Jun 25, 2009
    Location:
    Home of the Birmingham Civil Rights Institute
    #3

    There is zero chance of Apple using the high end E5s - TDP is too high and Apple and Intel tax would gag.
    Geekbench score of only 41,000 for a machine costing that much money is just satisfactory, but not great. If Apple decides to make this party, it'll be adorned in, at best, dual < 2.6 GHz diamonds.
     
  4. wallysb01 macrumors 65816

    Joined:
    Jun 30, 2011
    #4
    Presumably if one desired they could buy the base DP and put those E5-2690s in there themselves right? People are doing that with the X5690 and the TDP is not substantially different. At $2000 a pop though, I don't think too many would be doing such a thing.
     
  5. Tutor, Apr 6, 2012
    Last edited: Apr 6, 2012

    Tutor macrumors 65816

    Tutor

    Joined:
    Jun 25, 2009
    Location:
    Home of the Birmingham Civil Rights Institute
    #5
    If Apple decide to grace the Sandy Bridge party, you're 100% right.


    You're 92 to 110% right because it would cost about $800 more than swapping in the X5690 which aren't cheap, particularly when you start with having already bought the base system (you're 100% right so far) . But the base price needs to be the lowest possible, so I think it would be best to wait until the base systems start appearing for sale as refurbs, if you want to go that route (this is why a gave a range from 92% to account for starting from the lowest base price possible based on a guess about the retail/refurb price differences). You get 110% because I suspect that used versions of Sandy Bridge E5s will garner a relatively lower resale price (compared to 5500s and 5600s) because the market for the slower version of these chips may well be smaller because unlike the Nehalems and Westmeres which could be clock to yield up to 60% greater performance, the current Sandy Bridge Xeons can be clocked to yield only about 5% greater performance. They are locked down in almost every way. Thus, tweakers aren't that enthusiastic about them like they were about the earlier low end Xeons. So your resale price recovery may be lower.
     
  6. wallysb01 macrumors 65816

    Joined:
    Jun 30, 2011
    #6
    Sounds fair.
     
  7. Tutor macrumors 65816

    Tutor

    Joined:
    Jun 25, 2009
    Location:
    Home of the Birmingham Civil Rights Institute
    #7
    You may be entitled to a 110% the more I think about it.
     
  8. goMac macrumors 603

    Joined:
    Apr 15, 2004
    #8
    Eh, the TDP is a good point, but the price for performance isn't something I agree with.

    The amount you pay for more performance is on an exponential curve, not a linear curve. So I would expect to pay 4x for a machine 2x as fast.
     
  9. Tutor macrumors 65816

    Tutor

    Joined:
    Jun 25, 2009
    Location:
    Home of the Birmingham Civil Rights Institute
    #9
    In theory, we might not have so much of a disagreement if one is not a tweaker and if Apple were to market the top of the line E5 chips in Mac Pros and if Apple reduced the Apple tax. Then one would not have to go through the gyrations and unnecessary low end chip/system purchase to get to swap out the base chips to get top of the line E5 performance. Furthermore, a Geekbench 2 score of 41,000 is not twice as fast as that of the current top of the line Mac Pro. The average Geekbench 2 score for the top of the line Mac Pro is 24,159 [ http://www.primatelabs.ca/geekbench/mac-benchmarks/#64bit ]. So 41,000/24,159 = 1.70. But, in fact, what Apple will market will not be 1.7x as fast - it won't even be 1.5x as fast. If Apple attends the party, most likely it'll be with the E5-2670, with a base speed of 2.6 GHz and a turbo base of 3.0 GHz and a max of 3.3 GHz. The top of the line E5s have base speeds of 2.9 GHz (E5-2690) and 3.1 GHz (E5-2687W), but much higher frequency turbo ranges (3.3 GHz to 3.8 GHz and 3.4 GHz to 3.8 GHz, respectively. See post #11 at [ http://forums.macrumors.com/showthread.php?t=1333421 ]). Thus, Apple's, top of the line will be 10-15 % slower than the top of the line E5s. In fact, it'll be beyond the expectations of most if it is 25% faster in real world applications. "The average performance increase, according to IXBT Labs and Semi Accurate as well as many other benchmarking sites, at clock to clock is 11.3% Average compared to the Nehalem Generation, which includes Bloomfield,Clarkdale and Lynnfield processors" [ http://en.wikipedia.org/wiki/Sandy_Bridge ]. So if the 2012 dual 2.6 GHz Mac Pro were to be even 15% faster than a 2010 dual 2.93 Ghz Mac Pro, then that might be joyous. And it's just that my underclocked X5680s perform at a comparable benchmark level to the top of the line E5s, i.e., my WolfPack1's Geekbench 2 score is over 40,000 and I built it in one evening for less than the price of the current top of the line Mac Pro comparing the same options. Moreover, I've had it for almost a year and a half. Apple's top of the line 2012 Sandy Bridge Mac Pro, if there be any, will not be as fast as my old system - that I can guarantee. Thus, that's the basis for my statement, "Geekbench score of only 41,000 for a machine costing that much money is just satisfactory, but not great."
     
  10. jshbckr macrumors 6502

    Joined:
    Apr 20, 2007
    Location:
    Minneapolis, MN
    #10
    You say this when the post clearly shows the current top Mac Pro score as nearly double the score of the top iMac.

    My Mac Pro has scored over 11,000 on geekbench and it's now 5 years old. It'll be another few years before iMacs reach the current 24,000+ Mac Pro score.
     
  11. Umbongo macrumors 601

    Umbongo

    Joined:
    Sep 14, 2006
    Location:
    England
    #11
    the /s means sarcasm :)
     

Share This Page