New Barefeats tests show MacVidCards Titan-X the 5K champion, trounces D700s

Discussion in 'Mac Pro' started by MacVidCards, May 1, 2015.

  1. MacVidCards Suspended

    Joined:
    Nov 17, 2008
    Location:
    Hollywood, CA
    #1
    http://barefeats.com/mpt5k1.html

    The D700s were beaten by 100% in 3 of 6 tests. (Or you could say their score was 50% of the Titan-X)

    The 8 Core D700 did manage to come from behind and win one test, a Tessmark tessellation test at 16 levels. When the heat got turned up to 32 levels, the Titan-X took a 10% lead.

    I am rather surprised that the D700s managed to fare reasonably well in the few tests where they hadn't lost sight of cMP's taillights.

    I imagine that either PCIE 3.0 bus speed or having a 2nd D700 at hand helped there, one way or another.

    Oh, the iMac 5K was in the test too, it even beat the nMP in half of the tests.

    Those D700s are really showing their age.

    But when you see graphs like the attached, it makes all of the nMP apologists apologies seem especially hollow. The nMP was left for dead half way across the bar graph. So if you were running something that required OpenGl, why would you go with a machine that ran out of steam so easily?
     

    Attached Files:

  2. jetjaguar macrumors 68030

    jetjaguar

    Joined:
    Apr 6, 2009
    Location:
    somewhere
    #2
    I was toying with the idea of buying a classic mp and installing a 3690 and a 980 but not sure if it is worth it. Also .. can you even run an asus 4k monitor with it?
     
  3. MacVidCards thread starter Suspended

    Joined:
    Nov 17, 2008
    Location:
    Hollywood, CA
    #3
    Humbling defeat for nMP

    Do you mean the Asus PQ-321Q?

    I have a hack detailed here (thread with 1984 in title) and on our blog that gives full 60Hz on DP, provided you start up in HDMI and switch over to DP.

    In fact, when used with an SST 4K display, the 980 & Titan-X not only do 4K 60Hz natively, they give you a boot screen in 4K. Same on a 5K Dell.

    Used with a nMP, you will be staring at a black screen until desktop loads. Apple could fix that in a jiffy, if they weren't furiously coming up with "exciting" new watch bands.

    The cMP has easily surpassed the nMP in display abilities, thanks to old GPUs and neglect by our friends in Cupertino.
     
  4. jetjaguar macrumors 68030

    jetjaguar

    Joined:
    Apr 6, 2009
    Location:
    somewhere
    #4

    Yea that asus. MKBHD is running 2 of them with his nMP with no issues.
    I just don't know if it is worth dropping like 600 on a quad cMP then upgrading .. I need a desktop though
     
  5. dmylrea macrumors 68020

    dmylrea

    Joined:
    Sep 27, 2005
    #5
    Umm...because the majority of people looking for a powerful Mac for work aren't enthusiasts and know nothing about souping up old classic Mac Pro's with super expensive video cards. Instead, they walk into an Apple store, drink the Apple kool-aid and buy the most powerful item on the shelf. What do they have to compare it to? Something that runs Windows? It's not their fault.
     
  6. Surrat macrumors 6502

    Surrat

    Joined:
    Jun 20, 2014
    Location:
    United States
    #6
    Sadly, this is really the truth. I would guess that few mac buyers have any clue what the cMP is capable of, or how to hotrod it.
     
  7. linuxcooldude macrumors 68020

    Joined:
    Mar 1, 2010
    #7
    Enjoy it while it lasts. While the D700's may be showing their age, so are the classic Mac Pro's. We all know newer technology will be more faster and efficient then old ones.
     
  8. MacVidCards thread starter Suspended

    Joined:
    Nov 17, 2008
    Location:
    Hollywood, CA
    #8
    Yep, more faster is coming.
     
  9. Mattstrete macrumors member

    Joined:
    Nov 14, 2011
    #9
    I don't see what gaming benchmarks like these are supposed to show if the two D700s were not being run in crossfire, which is impossible in Mac OS. Barefeats has dealt with this issue only briefly in the past. So, technically, you're not comparing apples with apples.
     
  10. AndreeOnline macrumors 6502

    AndreeOnline

    Joined:
    Aug 15, 2014
    Location:
    Zürich
    #10
    Hopefully Rob has his OpenCL and pro app benchmarks coming up as well.

    These numbers are pretty lopsided, but if Diablo III is what you do then you have it black on white here.
     
  11. Mattstrete macrumors member

    Joined:
    Nov 14, 2011
    #11
    It's worse than that. The nMP is sold as a unit incorporating two graphics cards which are designed and intended to be run simultaneously, as is the case with certain "pro" apps in Mac OS and as is the case with Crossfire in Windows. To say that an "old" Mac Pro running the very latest desktop GPU "trounces D700s" and therefore the nMP, on tests where only ONE D700 was running, is not fair.
     
  12. mugwump macrumors regular

    Joined:
    Jan 10, 2004
    #12
    I agree that the nMP is certainly not a gaming rig. Any relevant test would be something like a video render test of some sort in Preimere that uses a variety of gpu acceleration and is fully multi core cpu.

    But I'm not sure about the actual implementation of the gpu's in the new Mac Pro. One is listed as a render gpu and the other is the OS, display driver gpu. Not sure if both combine for intensive tasks.
     
  13. nox-uk macrumors regular

    Joined:
    Apr 11, 2012
    #13
    Sorry to sound a little pedantic, but it doesn't trounce d700s as per your title/post. The results do indeed show it trouncing a single d700 but we knew it would anyway because it's somewhat of an epic videocard (with an epic price tag to match).

    Now if the barefeat's guy would post the crossfired results of a pair of d700s at 5k vs the titan, then you can shout it trounces d700s (which i'm pretty sure it will do for a lot of the tests anyway) or concentrating on the results of the OS X programs that utilise both D700's ?

    But by mentioning crossfire, we are heading into gaming territory now, which 99% of people aren't buying a mac pro for anyway.

    and does it work on your nMP as an eGPU yet? that's something I really want to see the results for :)

    Nox

    EDIT: Just seen Mattstrete's posts, saying basically the same heh - i read it early just before shopping and wanted to reply, whithout a page refresh when i got home :D
     
  14. avemestr macrumors regular

    Joined:
    Aug 14, 2012
    #14
    All the people in this thread pointing out, that the comparison is unfair, is of course totally right. Comparing a very expensive Titan-X to a single D700 is totally wrong.

    So, lets instead compare apples to apples. Lets compare two D700 (BTO option with a price of $600-1000) with two GTX 970 with a total price tag of $678 (probably cheaper some places)).

    Also, let the match take place on the D700s home turf: OpenCL.

    Luxmark Sala only GPUs:

    Current 8c nMP + 2 * D700: 3527 (Source: http://barefeats.com/tube20.html)
    6 year old 6c cMP + 2 x GTX 970: 4499 (Source: http://forums.macrumors.com/showthread.php?t=1840717)


    So current mid-tier GPUs with no support from Apple are 27% faster than the most expensive option from Apple in the area most heavily advertised by Apple.

    You are absolutely right. And that's good news for alle nMP owners! The GTX 970 will most certainly be outperformed by the latest after-market upgrade GPUs that Apple is just about to release for the nMP!
     
  15. MacVidCards thread starter Suspended

    Joined:
    Nov 17, 2008
    Location:
    Hollywood, CA
    #15
    You're right, it's totally unfair.

    I'll send him a 2nd Titan as soon as I can.

    But I have to say that I get a laugh out of hearing that the way to test a nMP fairly is to run Windows. If we're going there, shouldn't we get a current gen HP workstation and put 2 Titans in it? That should be a fun one. If Apple can't be bothered to implement Crossfire in its flagship product in it's own OS I don't think booting into someone else's OS makes a test "more fair".

    I imagine he will do a FCPX test, as "nMP is a FCPX console" is a fair assessment. But I really think that a 2nd Titan will do some serious OpenCl work that will make the nMP look bad on it's home field.

    And I think the "epic price" comment raises a rather important issue. How much are the D700s in retail as a separate purchase? Same price as a Titan-X in PC form. Even including our flashing fee any "bang for the buck" comparison also looks like a lopsided fight, again. Though I guess you could sell your D300s or D500s to recoup some, but then shouldn't the cMP side get to factor in the sale of whatever cards were in it? Trying to make this sort of test "fair" is a slippery slope.

    Rob plugged the machines into the wall and ran the tests. I think he might be publishing more results but I have no say in that.

    And if "trounce" is the wrong word, what else do you use to describe something that has double the frame rate in half the tests? How about a "rout"?
     
  16. bax2003, May 2, 2015
    Last edited: May 2, 2015

    bax2003 macrumors 6502a

    bax2003

    Joined:
    Dec 25, 2011
    Location:
    Belgrade, Serbia
    #16
    Everybody is missing one point about cMP and Dual GTX 970s....you need 4 x 6-pin PCIe power connectors for that combo, at least for reference design GTX 970. So you need external PSU or dedicated internal PSU 200-300w for 2nd GTX 970....and great....cMP power consumption rises to 1200w, that is just 3 x nMP. More noise and heat aside for now.
     
  17. AndreeOnline macrumors 6502

    AndreeOnline

    Joined:
    Aug 15, 2014
    Location:
    Zürich
    #17
    Well, a couple of us have modded the internal PSU to deliver two additional 8 pin PCIe cables allowing for two 8+6 cards. It's a bit involved, but not THAT hard.

    Actual mod takes less than 20 minutes.

    ----------

    It's quite promising that the 970s and newer (I assume) are performing well in OpenCL. Dual 970 seems to be great price/performance.

    I still think the driver situation is a bit shaky on the NVIDIA side, but... Nonetheless a great option.
     
  18. linuxcooldude macrumors 68020

    Joined:
    Mar 1, 2010
    #18
    While I couldn't say running two Titans would not be fair, most professionals are not going to want to mess around with extra power supplies and cables. Other then the enthusiasts who might congregate on these forums.

    Which comes around to the point having to hack the firmware/software/hardware. In order to make a classic Mac Pro run faster then what the new 2013 nMP and beyond, does not make for the best production machine people come to depend on in a daily basis to earn a living.

    Such as a blank screen with some Nvidia drivers breaking to only one monitor working with the GTX 570.
     
  19. Pressure macrumors 68040

    Pressure

    Joined:
    May 30, 2006
    Location:
    Denmark
    #19
    Just test something that uses FP64 ...

    I'm not saying the Titan X isn't awesome at gaming (it is).

    Double Precision performance is a joke on Maxwell but let's not highlight that fact.
     
  20. DonMega macrumors regular

    Joined:
    Jul 8, 2007
    #20
    What I think some people here don't realize is that MacVidCards wouldn't be hear trumpeting the cMP and his mods and bashing the nMP except for the fact that he is MAKING MONEY. :eek:

    That's what this all boils down to. His ability to market his products to people who think they can't afford a nMP or those who he has convinced that the nMP is an inferior product not worth having. Never mind the fact that he himself owns one. He has said he is working on an eGPU solution for the nMP. 2 years from now, maybe sooner, he'll be telling you all to buy a discounted nMP to make use of his latest eGPU hack and you'll be buying it all up.

    Ron Popeil would be proud. Consumerism at it's finest...
     
  21. MacVidCards thread starter Suspended

    Joined:
    Nov 17, 2008
    Location:
    Hollywood, CA
    #21
    OK, show us an OS X app that uses Double Precision.

    I looked for awhile, never found one.

    And if Double Precision is what you need, a Titan Black beats all comers, just a click on a checkbox and BANG...it's king.
     
  22. avemestr macrumors regular

    Joined:
    Aug 14, 2012
    #22
    Please stop spreading lies.

    If you look at the link, you'd notice that guy (and another one in the thread also running 2*970) did NOT have to do ANY hacks.

    They use ONLY the power present from the factory in the Mac Pro. No extra PSU. Also, the 970 has a TDP below 150W and the entire machine is therefore within limits of the PSU.

    Why do you write such nonsense?

    Steps to make cMP use faster GPUs than nMP:

    1) Install web drivers from NVIDIA.
    2) Insert card in slot.
    3) Boot machine.
    4) Enjoy.

    Is it really that hard?
     
  23. MacVidCards thread starter Suspended

    Joined:
    Nov 17, 2008
    Location:
    Hollywood, CA
    #23
    Point this out all you want, I have never denied that. But my motivations have nothing to do with those test results, they stand on their own. And the nMP looks rather silly at this point.
     
  24. flowrider macrumors 601

    flowrider

    Joined:
    Nov 23, 2012
    #24
    He owns one because he needs it for development purposes. Many of use here own his modified products. Not because he brainwashed us, but because we have made informed decisions to retain the cMP and modify and/or expand it's capabilities.

    You are free to make your own decisions! BTW, I notice you have a cMP that appears to be modified.

    Lou
     
  25. netkas macrumors 65816

    Joined:
    Oct 2, 2007
    #25
    FP64 tests in osx were and are always great, for dickmetering.
     

Share This Page