1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Graphics nVidia vs ATi - DX11 - finally we know the winner?

Discussion in 'Hardware' started by Synay, 5 Jul 2010.

  1. Synay

    Synay What's a Dremel?

    Joined:
    25 Oct 2009
    Posts:
    442
    Likes Received:
    7
    If anyone had any doubt about what side to choose for a high-end gaming rig, I think that will sort out that problem nicely. ATI fans, you might not want to watch that :)

     
  2. Xonar

    Xonar What's a Dremel?

    Joined:
    2 Nov 2009
    Posts:
    676
    Likes Received:
    33
    Completely synthetic benchmark which gives very little indication of true in-game performance, pointless video tbh.
     
  3. Jipa

    Jipa Avoiding the "I guess.." since 2004

    Joined:
    5 Feb 2004
    Posts:
    6,363
    Likes Received:
    125
    What a stupid test. It's a benchmark and the tessellation in that is so hugely over-done it doesn't really even try to represent current games.

    What ever.
     
  4. 13eightyfour

    13eightyfour Formerly Titanium Angel

    Joined:
    9 Sep 2003
    Posts:
    3,430
    Likes Received:
    131
    THIS.

    Personally i prefer nvidia cards as i find their drivers much better than ATis, but just look at the power usage! The nvidia setup uses more power over the ATi cards than the size of my PSU.

    If your building an ultra ultra high end gaming computer then yeah 4x GTX480 is probably the best option. But If it showed real world benchmarks the results wouldnt be so conclusive imo
     
  5. sleepygamer

    sleepygamer More Metal Than Thou

    Joined:
    24 Apr 2010
    Posts:
    1,064
    Likes Received:
    72
    As desirable as that kind of power is, look at the power draw... and the price!

    ...still, I'll try persuade the girl.
     
  6. roosauce

    roosauce Looking for xmas projects??

    Joined:
    12 Feb 2010
    Posts:
    930
    Likes Received:
    47
    Did you post a video link? Cant see it.

    I presume you're talking a Heaven benchmark, which others have noted is synthetic and not really true to life. I dont think that we have seen any really good application of dx11 yet, so the jury is out.

    I think that it will take a die shrink for nvidia's cards to appeal to me however. I was tempted by the 480 at last upgrade but it is just too cumbersome.
     
  7. shoxicwaste

    shoxicwaste Minimodder

    Joined:
    2 Jul 2010
    Posts:
    248
    Likes Received:
    9
  8. Bloody_Pete

    Bloody_Pete Technophile

    Joined:
    11 Aug 2008
    Posts:
    7,701
    Likes Received:
    634
    Why is everyone acting suprised that the 480's are faster than 5870's? For the cost I'd want it to be faster than it is. I personally would have prefered it to be compared to a 5970, as its the ATI card that's level with the 480 in price, heat and power usage.
     
  9. azrael-

    azrael- I'm special...

    Joined:
    18 May 2008
    Posts:
    3,852
    Likes Received:
    124
    As abundantly mentioned the Heaven benchmark is far removed from real-world applications. Also, there's that little fact that it's an EVGA-conducted benchmark as per the title of the YouTube video.
     
  10. wyx087

    wyx087 Homeworld 3 is happening!!

    Joined:
    15 Aug 2007
    Posts:
    11,097
    Likes Received:
    363
    so, OP, who is the winner?

    performance: nVidia
    efficiency: ATI
    which one do you consider to be winner?

    why do ATI fans want to look away? is it because ATI cards use so little power?
     
  11. Pete J

    Pete J Employed scum

    Joined:
    28 Sep 2009
    Posts:
    6,245
    Likes Received:
    820
    Sounds like a couple of butthurt ATI fanbois :naughty:

    I joke, I joke! Please don't take it badly :)

    I'll try to weigh up things but I know that I'm biased.

    1) ATI cards don't seem to scale as well as Nvidia cards and with four way set ups, this is going to be greatly exaggerated. Personally, I wouldn't be surprised if three 5870s performed better than four.

    2) I have my doubts that four 480s stuck together like that would be able to play the demo in a continuous loop for long. On my cards I've seen temperatures of 102C (admittedly overclocked) recorded after a day of gaming. I'm willing to bet a Furmark run would make them fail.

    3) Here's a controversial one: I genuinely believe that the Heaven benchmark is indicative of future DX11 titles and as such look forward to an Nvidia dominance in performance for these aforementioned titles.

    All I know is that I'm currently damn happy with my purchases at the moment - as I imagine anyone who has a 5770, 5850, 5870 or 5970 is.
     
  12. sleepygamer

    sleepygamer More Metal Than Thou

    Joined:
    24 Apr 2010
    Posts:
    1,064
    Likes Received:
    72
    No doubt. If I were to run any multi GPU setup with cards that close, I would move right to 3rd party coolers, or watercooling.

    Although putting the four GTX 480 cards together and utilising the heatsinks as a sort of crude pan, I think that much joy could be had in making a combination gaming/breakfast machine.
     
  13. roosauce

    roosauce Looking for xmas projects??

    Joined:
    12 Feb 2010
    Posts:
    930
    Likes Received:
    47
    Lol - I see the video now ... I was on my iphone ...

    Could be that this is true ... more fun would be unlimited detail.

    I think we've covered that super-heavy tesselation isn't useful just yet. I think there's some time - both for nvidia to use a smaller process, and for ATI to re-design their tesselator - before this future comes about.

    We've had a look at the super-high-end £1600 worth of GPU ... but I hear that the ATI tesselation unit in cypress cards basically remains the same across the range - the 5770 has the same unit as the 5870. I wonder what the comparison is like in the mid-range volume market - GTX 460s (well 465 for now ...) and 5770s? Anyone know?
     
    Last edited: 5 Jul 2010
  14. Rofl_Waffle

    Rofl_Waffle What's a Dremel?

    Joined:
    24 Mar 2010
    Posts:
    504
    Likes Received:
    12
    That is pretty much true. By the time this level of tesselation comes in the form of games, both companies would have something new and different.

    For the time being, in a real game the difference is less pronounced though the power consumption and heat is fairly accurate. Nvidia has the performance crown but ATI has better bang for the buck. But with the horrible scaling of 4 GPUs, nobody can say that was money well spent. Just a lot of heat produced. 1000W is more than an actual space heater. 1500W is like a toaster oven. Having that on 24/7 is going to spike your AC bill.
     
  15. Synay

    Synay What's a Dremel?

    Joined:
    25 Oct 2009
    Posts:
    442
    Likes Received:
    7
    That all may be true, including the wattage but then again, ATI produces stuttering FPS at a 1000 watts and Nvidia doubles the frame rate at 400 watts more. I don't know how about you but I would rather have a hugely playable rates with 1400W rig than unplayable for 1000W.
     
  16. Bakes

    Bakes What's a Dremel?

    Joined:
    4 Jun 2010
    Posts:
    886
    Likes Received:
    17
    Yes, especially since this is a a synthetic benchmark which has likely been optimized for by nVidia and ATI - scaling in this won't give the results people hope for.

    I think that tessellation will be a big thing in the future, but it won't happen to the levels that Heaven 2.0 takes it.

    For example, here is a Heaven wireframe:

    [​IMG]

    Yes, there is more detail in the tessellated wireframe, but much of the extra detail could just be added by putting more exaggerated features in the source. If you see the number of (basically) useless polygons added by tessellation, it's pretty obvious that Heaven is overdoing it for effect.

    So am I :D
     
  17. Slizza

    Slizza beautiful to demons

    Joined:
    23 Apr 2009
    Posts:
    1,738
    Likes Received:
    120
    We already knew SLI scaling is better than xfire.
     
  18. roosauce

    roosauce Looking for xmas projects??

    Joined:
    12 Feb 2010
    Posts:
    930
    Likes Received:
    47
    Yes but QUAD??? I really don't see the point of using more than 2 high-end cards - except for big benchmark numbers. I'm not at all upset at AMD for not scaling well at 3-4 cards. Nvidia has done very well to get 80-90% scaling at 2 cards however - that's very cool.
     
  19. Bloody_Pete

    Bloody_Pete Technophile

    Joined:
    11 Aug 2008
    Posts:
    7,701
    Likes Received:
    634
    This may claify things (or could just be me making stuff up):

    Performance Cost Power Total
    ATI 39.8 1280 950 0.017847534
    Nvidia 81.4 1760 1500 0.024969325


    Using Performance/(cost+power)

    The Nvidia total is higher, so although it costs more and uses more power the performance is still very high.

    Kind of makes sense to me. But I'd still like to see 4 5970's do that just to see what they'd get...
     
  20. Slizza

    Slizza beautiful to demons

    Joined:
    23 Apr 2009
    Posts:
    1,738
    Likes Received:
    120
    I don';t think you can run 4x HD5970 cards..
    Limit was 4gpus last i knew.
    I may be wrong.
     

Share This Page