Cinebench R15 on the nMP?

Discussion in 'Mac Pro' started by ScottishCaptain, Dec 22, 2013.

  1. ScottishCaptain macrumors 6502a

    Joined:
    Oct 4, 2008
    #1
    Does anyone have any real scores from this sucker for the new Mac Pro?

    The Macworld crap that got posted is worthless, since the time to completion means absolutely nothing (I'm guessing that includes the time it takes to load assets and prepare them in memory before the benchmark has even run- what a bunch of idiots). And I'm getting tired of seeing "Geekbench" scores flung all over like they actually mean something.

    I see some people have witnessed these machines in person already at an Apple store. Is there any chance someone would be willing to take down a copy of Cinebench R15 on a USB flash key, run it, and post the results here so we can see what we're dealing with?

    -SC
     
  2. echoout macrumors 6502a

    Joined:
    Aug 15, 2007
    Location:
    Austin, Texas
    #2
    I thought that was weird too. Why measure in seconds when it spits out a score that is measurable against others on numerous databases? Pretty strange.
     
  3. ScottishCaptain thread starter macrumors 6502a

    Joined:
    Oct 4, 2008
    #3
    It's not weird, it's downright suspicious. Whoever did that had to find a stopwatch and time it themselves, because Cinebench won't give you that information.

    Furthermore, CB tends to take a few seconds to start rendering (just as Cinema 4D does). The amount of time it takes between pressing the run button and the test actually starting seems to vary wildly depending on other (pointless) aspects of the system. My crappy Macbook Pro starts the test almost immediately, compared to my Mac Pro- which sits around for a good 20 seconds before anything happens (even though the Mac Pro kicks the pants off the laptop, as it should).

    Hence my desire to see ACTUAL Cinebench R15 results of the new Mac Pro, irregardless of how flattering or not they might actually be. I'm getting really tired of this inflated numbers ******** game and dance surrounding the computer. I just want to see some actual data that I can use to make an informed purchase decision.

    -SC
     
  4. linuxcooldude macrumors 68020

    Joined:
    Mar 1, 2010
    #4
    Some are saying Cinebench is not that accurate either, its cpu bound.
     
  5. chris.k macrumors member

    Joined:
    May 22, 2013
    Location:
    YSSY
    #5
    I'll try to sneak back into the Sydney Apple store and run the benchmark for ya tomorrow.

    - CK.
     
  6. deppest macrumors member

    Joined:
    Oct 6, 2009
    #6
    It's not CB R15 but I came across this CB R11 CPU value on a German site:

    test config: Mac Pro 8-core Xeon E5-CPU (3,0 GHz), dual D700, 64 GByte RAM (DDR3-1867 ECC), 1 TByte SSD

    CB R11 CPU: 13.61


    Not very impressive compared to MP 2009/2010 (my config below scores about 12) , but clearly the nMP has not been designed with primarily CPU-bound rendering tasks in mind. Let's at least hope it will push for some more wide-spread implementation of OpenCL in rendering.


    http://dark-network.org/showthread.php?30546-Apples-neuer-Mac-Pro-im-Kurztest&p=66146
     
  7. theSeb macrumors 604

    theSeb

    Joined:
    Aug 10, 2010
    Location:
    Poole, England
    #7
    For comparison

    [​IMG]
     
  8. echoout macrumors 6502a

    Joined:
    Aug 15, 2007
    Location:
    Austin, Texas
    #8
    Cinema 4D is my most used application on my Mac Pro, so this is helpful. Pretty sad scores.
     
  9. fig macrumors 6502a

    fig

    Joined:
    Jun 13, 2012
    Location:
    Austin, TX
    #9
    I'd expect more, but this is also a realllly small sample that may or may not be accurate so I'll wait to see some more data before passing judgement. The numbers on the Mac Pro site look promising but their 1.0 "baseline" doesn't have any reference point that I can find.

    A Mac Pro for me will be primarily to run Maya and Modo so I'm hoping for strong 3D performance. Given that Modo has been used in some recent demos and benchmarks for the Mac my fingers are crossed...
     
  10. Tutor, Dec 23, 2013
    Last edited: Dec 25, 2013

    Tutor macrumors 65816

    Tutor

    Joined:
    Jun 25, 2009
    Location:
    Home of the Birmingham Civil Rights Institute
    #10


    No real scores yet. However. MMV greatly w/nMP, depending on how Apple deals with thermals. I have an < $9K 4 CPU self-build, running four E5-4650s {version QBED} that scores 3,791 in CB 15 now that I've figured out how to keep the CPUs cooler. Before that it scored only about 3,600. An early Geekbench 2 score posted by Apple seemed to indicate to me that the 12-core was throttling. Throttling will decrease Cinema 4d and Cinebench performance.
     
  11. sirio76, Dec 24, 2013
    Last edited: Dec 25, 2013

    sirio76 macrumors regular

    Joined:
    Mar 28, 2013
    #11
    Tutor nicely canceled his post, anyway I leave my reply..
    Tutor, the 9k price tag for a 4 xeon system is a bit misleading. For 4x e5-4650 the cost is ~14500$according to intel: http://ark.intel.com/products/64622/
    On amazon they sell for about 5k each even refurbished they sell for 4k, but let's say you pay only 14.500$. Add all the rest and a 4 way system will be likely around 20k(even if you assemble everything by yourself), unless you buy everything used. So real price/performance ratio is 20k / CB15 score of ~3800.
    For just 17k I can buy a fully loaded 8core nMP(~7k, probably around a CB15 score of ~1400), add a small farm of 10 6core/12t render nodes(CB15 score 1200, <300W, 1000$each), in total 12000+1400=13400, even after a loss in performance due to network latency you will get an overall result of >11000 CB score, I'll have much more render power for less money than a 4way system, and I'll have also better single score performance(very important in Cinema4D) due to 3.9ghz turbo of the nMP. I can already match a 3800 result even with just two render nodes(1000$ yeach) and an 8core, so 9k, but everything new. Of course I can get everything cheaper if I buy used stuff.
     
  12. Tutor, Dec 24, 2013
    Last edited: Dec 25, 2013

    Tutor macrumors 65816

    Tutor

    Joined:
    Jun 25, 2009
    Location:
    Home of the Birmingham Civil Rights Institute
    #12


    That price isn't misleading. It's what I paid for them used - everything else is new. If, like you say, "... they sell for about 5k each [sic] even refurbished they sell for 4k," then maybe you should look on E-bay where I routinely see them used for $500 or less. Just google, " Ebay E5-4650 QBED." These CPUs do need active coolers (with fans) because they are clocked for turbo boosting like the E5-2680 V1 3.5 GHz vs. 3.3 GHz for the final production 4650s. That the CPUs are used doesn't bother me because throughout my like I've learned to carefully select used items that maximize performance. Also, since I use C4D, upon which Cinebench is based and emulates, keeping the CPUs cool is very important for long renders. It also helps to get higher Cinebench performance. That's why I posit that the nMP's Cinebench 15 scores will be dependent, in part, on how well the nMP handles thermals, particularly in the case of those CPUs that run, at base, at 3.5 GHz of more.
     
  13. sirio76, Dec 24, 2013
    Last edited: Dec 25, 2013

    sirio76 macrumors regular

    Joined:
    Mar 28, 2013
    #13
    Tutor nicely canceled his post, I leave my reply.
    Thank's for the story of your life. And thanks also for illuminating us with your knowledge and wisdom.
    That being said, if I buy a 10k computer for just 1k because of a very good deal or because my uncle works for Intel, that doesn't mean that everybody can easily do the same or want to buy used stuff.
    Is that simple, if I want to purchase a new 4x xeon system tomorrow, I'm going to spend much more than 9k.
    Nothing wrong with used stuff, but for comparison purpose it's better to compare "new vs new".
    About the "theory vs already", well that's not theory, it's 2+2=4, and if I buy my computer in a couple months, it will always be 2+2=4. Even now(and by now I mean I'm rendering in C4D right in this moment) I've an overall CB15 score of ~3500 with just two cheap 6core nodes and two 6 year old 8core macpro rendering together via Vray DR(or Team render).
     
  14. Tutor, Dec 24, 2013
    Last edited: Dec 25, 2013

    Tutor macrumors 65816

    Tutor

    Joined:
    Jun 25, 2009
    Location:
    Home of the Birmingham Civil Rights Institute
    #14
    Nice render farm you have there - it's much better than what I started with. If your uncle works for Intel then he probably can answer the central question or he may know someone who can provide better insights into how well the CPUs in the nMP perform relative to earlier ones to better extrapolate how the nMP will perform under Cinebench 15.
     
  15. ScottishCaptain thread starter macrumors 6502a

    Joined:
    Oct 4, 2008
    #15
    Can we please get back on topic?

    No offence Tutor, but I really don't care about what you may or may not be running and how much it cost you. That is not the point of this thread.

    The point of this thread was to try and get Cinebench R15 scores from the new Mac Pro so that we would have a valid comparison point to make educated purchase decisions. You have borderline hijacked this thread and made it about your systems, which I couldn't care less about.

    -SC
     
  16. sirio76, Dec 24, 2013
    Last edited: Dec 24, 2013

    sirio76 macrumors regular

    Joined:
    Mar 28, 2013
    #16
    The 8core nMP should be nearly as fast as a 12core 2,93ghz according to other benchmark tools, I'm expecting 1300/1400 CB15 scores. The 6core should reach about 1000 points(the i7 4930k I use in my nodes are basically the same processor and at stock speed before OC they reach 950 points). Don't know about the 4 or 12 core version.
    There was one of the many crappy reviews reporting CB score, but honestly their tests seems rubbish from clueless people.
     
  17. Tutor, Dec 24, 2013
    Last edited: Dec 25, 2013

    Tutor macrumors 65816

    Tutor

    Joined:
    Jun 25, 2009
    Location:
    Home of the Birmingham Civil Rights Institute
    #17
    I project that the 4-core CPU will score on average around 1,010 and the 6-core CPU will score around 1,095, but how well those CPUs perform in the nMP can only be settled shortly after mass shipment.
     
  18. sirio76 macrumors regular

    Joined:
    Mar 28, 2013
    #18
    Don't you think is a little bit strange that 4 and 6 core give essentially the same results according to your numbers(1010/1095)?
    I agree on the 6core result, but the 4core should be more like >800.
    The correct link is www.cbscores.com
     
  19. Tutor, Dec 25, 2013
    Last edited: Dec 25, 2013

    Tutor macrumors 65816

    Tutor

    Joined:
    Jun 25, 2009
    Location:
    Home of the Birmingham Civil Rights Institute
    #19
    I give a little more weight to core speed + I hope that the 4-core Xeon CPU clocked at 3.7 GHz is actually faster than a 4-core i7 4770K clocked at 3.51 GHz www.cbscores.com. But in the end, time will tell.
     
  20. RAILhead macrumors newbie

    Joined:
    Aug 5, 2002
    Location:
    Texas
    #20
    I would like to see this, too. I'm currently running a late 2012 27" iMac, 3.4GHz 4 core i7, 32GB RAM, GTX 680MX. I've been doing a LOT of Final Cut Pro X, HandBrake, and MediaConverter work (along with tons of Photoshop, Aperture, etc.). I would LOVE to have my video work time cut down, but not sure how far to go with the nMP. CB15 gives me:

    OpenGL: 64.82 fps
    CPU: 661 cb
    CPU (single core): 137 cb
    MP Ratio: 4.82x

    Some say the 6 core is the sweet spot...
     
  21. sirio76, Dec 25, 2013
    Last edited: Dec 25, 2013

    sirio76 macrumors regular

    Joined:
    Mar 28, 2013
    #21
    Those 4770k are over clocked @4.3/4.5ghz(Cinebench do not report correctly clock speed), so they are not very good for this comparison.
    http://forums.cgsociety.org/showpost.php?p=7672743&postcount=52
    On another thread an early owner report a score of about 970 point for a nMP 6 core, that unsurprisingly give nearly the same result of an i7 4930k @stock speed.
    I hope to see some accurate results for the 8/12 core model soon, and I hope there will not be thermal related problems(much more important than a Cinebench test).
    Merry Christmas:)
     
  22. Tutor, Dec 26, 2013
    Last edited: Dec 26, 2013

    Tutor macrumors 65816

    Tutor

    Joined:
    Jun 25, 2009
    Location:
    Home of the Birmingham Civil Rights Institute
    #22
    Thanks for update and site reference - What does the pic show at the referenced site?

    Happy New Year - may it be your best one yet.
     
  23. sirio76 macrumors regular

    Joined:
    Mar 28, 2013
    #23
    Thank you my friend;)
    The pictures just show a cinebench result of 860points running the 4770k @4.5ghz(sorry, cgsociety attachments are only for members)
    Little off topic.. is there any place where I can see some images renderer with your powerfull systems?
     
  24. Tutor macrumors 65816

    Tutor

    Joined:
    Jun 25, 2009
    Location:
    Home of the Birmingham Civil Rights Institute
    #24
    Thank you again. Now I see more fully what you meant. Keep the info flowing.

    I haven't posted any images on the internet yet. But I plan to begin posting images shortly. I'll send to you a private message with the URL when I do so. Please reciprocate since we both use C4d, and I look forward to you enlightening me more about using Vray with C4D.:)
     

Share This Page