Crysis 3 VGA Graphics Benchmark performance review

Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Feb 22, 2013.

  1. Mufflore

    Mufflore Ancient Guru

    Messages:
    11,413
    Likes Received:
    611
    GPU:
    Redacted by NDA
    None of those CPUs are overclocked.

    He uses AA.


    My tests have been over specific areas so far, no the whole game or a worst case scenario.
    But you can at least compare with the areas I have benched with.
     
  2. Evasive

    Evasive Member

    Messages:
    18
    Likes Received:
    0
    GPU:
    EVGA 580 GTX
    So to show a review of the best single GPU Card in the world, They were worried about that the crappier card would do in framerate?

    Does that make any sense?....Who cares what the crappier cards show.

    Edit: Like to add, in Crysis 3, they only use FXAA. No MSAA+FXAA results were shown.

    Edit #2: In the titan review, With HItman and FC3, only 2xMSAA was shown, there was no 4xMSAA or 8xMSAA results.
     
    Last edited: Feb 24, 2013
  3. WhiteLightning

    WhiteLightning Don Illuminati Staff Member

    Messages:
    28,115
    Likes Received:
    1,038
    GPU:
    GTX1070 iChillx4
    he uses FXAA and gets 29fps on a gtx580 at higher res then your running 1920x1200

    [​IMG]
     
  4. Mufflore

    Mufflore Ancient Guru

    Messages:
    11,413
    Likes Received:
    611
    GPU:
    Redacted by NDA
    They gave you a chart showing all the AA modes performance.
    You might want to do your own review, its a lot of work.
     

  5. Evasive

    Evasive Member

    Messages:
    18
    Likes Received:
    0
    GPU:
    EVGA 580 GTX
    It is alot of work, which is why I usually come to guru3d for my reviews.

    Please don't try to make stupid remarks. I am being totally honest, and I am asking honest questions. No trolling, just an observation.
     
  6. Mufflore

    Mufflore Ancient Guru

    Messages:
    11,413
    Likes Received:
    611
    GPU:
    Redacted by NDA
    Hmm, it does seem odd.
    My gfx is overclocked almost 20%, but that doesnt explain it enough I dont think.

    I just tried FXAA and managed to get 112fps when the tower exploded, it was still around the 110fps mark.

    I put blur on highest to see if that has an impact, made no difference.

    OK, I'll get fraps installed and will do the same test run as HH did.
    I've never used fraps to do an average framerate test, is there anything I need to know?
     
  7. Mufflore

    Mufflore Ancient Guru

    Messages:
    11,413
    Likes Received:
    611
    GPU:
    Redacted by NDA
    Hahaha, you wont believe the issue?!
    Running Fraps and MSI Afterburner both with OSD on, Fraps shows exactly half the framerate of Afterburner.

    Because 60fps with Afterburner looks very smooth, this seems to be a problem with Fraps !!
     
  8. WhiteLightning

    WhiteLightning Don Illuminati Staff Member

    Messages:
    28,115
    Likes Received:
    1,038
    GPU:
    GTX1070 iChillx4
    Sounds more reasonable that the issue is with afterburner tbh.
     
  9. Mufflore

    Mufflore Ancient Guru

    Messages:
    11,413
    Likes Received:
    611
    GPU:
    Redacted by NDA
    Yep you are right.
    I set Afterburners frame limiter to 2fps and it updates the screen every second.

    Well damn, it looks good for 30fps lol.

    ps I'm running Afterburner V2.21, its not that recent so the newer version may be ok.
     
    Last edited: Feb 24, 2013
  10. Hilbert Hagedoorn

    Hilbert Hagedoorn Don Vito Corleone Staff Member

    Messages:
    36,460
    Likes Received:
    5,487
    GPU:
    AMD | NVIDIA
    Click:

    http://www.guru3d.com/articles_pages/crysis_3_graphics_performance_review_benchmark,2.html

    Second chart, as written, that GeForce GXT Titan at work. Now correct me if I am wrong, but doesn't that show 9 different AA modes at three different resolutions = 27 AA tests for Titan alone ?

    Gentlemen, with VGA performance reviews it is important to find an image quality settings that works out for at least mid-range to top.

    Not everybody agrees on FXAA, however it's the best compromise as it utilizes low amounts of graphics memory whilst offering good performance. MSAA will immediately have an adverse effect, however ... MSAA does have better AA quality.

    With these articles you need to make a compromise, I opted FXAA as it's cheap AA whilst the graphics card overall have a really hard time to render the game. So you want 'cheap AA'. If not and you have the capability, simply flick on another AA mode, honestly good for you.

    For this review I opted to test HIGH quality and Very high quality. With the time at hand (and mind you that I was still working on this article Friday night at 11PM) these are the results that are offered.

    In a perfect world we'd test every AA mode available with every image quality settings. But guys. we tested 22 cards x 3 resolutions x 2 image quality modes = 132 test runs. And that's without all the other additional tests.

    We simply need to make choices. Also everybody has a different opinion. I remember doing a VGA perf article where we tested MSAA only and got gutted why we didn't test FXAA. So whatever we try to do .. there will always be compromises and unfortunately also always people that complain about our free content that I do in my free time.

    So be it.
     

  11. txd1979

    txd1979 New Member

    Messages:
    6
    Likes Received:
    0
    GPU:
    Gigabyte GTX 980Ti
    Are there any plans of updating the article with CPU-scaling in the future Hilbert?
    (Sorry if this has already been answered, but I read quick through the 6 pages of this thread, and couldn't find any info about it)
     
  12. Hilbert Hagedoorn

    Hilbert Hagedoorn Don Vito Corleone Staff Member

    Messages:
    36,460
    Likes Received:
    5,487
    GPU:
    AMD | NVIDIA
    No I'm done with all the comments, I will not be spending more time on VGA perf reviews. Whatever we do it's never enough.
     
  13. txd1979

    txd1979 New Member

    Messages:
    6
    Likes Received:
    0
    GPU:
    Gigabyte GTX 980Ti
    Don't get me wrong. I'm very happy with the article ;)
     
  14. Matt26LFC

    Matt26LFC Ancient Guru

    Messages:
    3,061
    Likes Received:
    19
    GPU:
    HD630 :(
    Ignore the idiots Hilbert, theres more than enough info in your reviews! People need to understand that you can't test every single possible hardware combination going and test every little thing they want, your only one man and already do an exceptional job!
     
  15. BangTail

    BangTail Ancient Guru

    Messages:
    3,590
    Likes Received:
    1,126
    GPU:
    EVGA 2080 Ti XC
    I totally agree with your methodology tbh.

    You aren't playing C3 with 8XMSAA on anything available today and getting a playable experience (Maybe @ 1024 x 768 ;) ).

    Sure, 8XMSAA looks better but you're giving up half your performance and I'm playing the game, not stopping to look at the edges on every object.

    I am running at 1600P which helps with edges on it's own but I've always found FXAA/SMAA to be an acceptable compromise.

    You get your smoothing and you don't lose any significant performance.

    Win/win AFAIAC.

    If the overhead exists to use better forms of AA, fantastic! In the case of C3, it is already extremely demanding so the ~50% hit of MSAA is not something I am willing to live with.

    I'm still dipping into the low 20s in some areas and that is just with FXAA enabled.

    Obviously, this is user specific but as others have said, the guy can only do so many tests.
     
    Last edited: Feb 24, 2013

  16. Evasive

    Evasive Member

    Messages:
    18
    Likes Received:
    0
    GPU:
    EVGA 580 GTX
    thank you for your time to respond. I really appreciate it.

    I know benchmarking sucks ass....specially with everything you must run resolutions etc etc etc.

    I might disagree with some of your reasons as to why you bench the way you did, but the time and effort you put into is so massive, so yea I understand you cant please everyone.

    Thanks again for your time HH, and keep the reviews coming
     
  17. txd1979

    txd1979 New Member

    Messages:
    6
    Likes Received:
    0
    GPU:
    Gigabyte GTX 980Ti
    I don't know why I got misunderstood here, but I wasn't complaining about the article. I'm extremely happy with Hilberts work, and it was just a simple question because there was a page about CPU-scaling in the Far Cry 3 test.
    I'm sorry if that made me look like I was complaining :(
     
  18. Hilbert Hagedoorn

    Hilbert Hagedoorn Don Vito Corleone Staff Member

    Messages:
    36,460
    Likes Received:
    5,487
    GPU:
    AMD | NVIDIA
    Don't worry that was not related to you buddy.
     
  19. alanm

    alanm Ancient Guru

    Messages:
    8,977
    Likes Received:
    1,326
    GPU:
    Asus 2080 Dual OC
    Welcome to the forum :). I think people can and should critique reviews as it can help improve future ones. But sometimes an odd abuser may show up and throw warrant-less claims of bias or improper test conditions or whatever while insulting the admin (as we saw yesterday). And after the great effort put into these reviews, I can imagine anyone not wanting to put up with it. So maybe your input was initially viewed warily by some in light of that. But dont worry about it, I'm sure everyone means well. Again, welcome to the forum.
     
  20. Noisiv

    Noisiv Ancient Guru

    Messages:
    6,657
    Likes Received:
    499
    GPU:
    2070 Super
    MSAA is subpar in most of deferred engines. Crysis 3 is no exception.
    To each its own, but I really don't see why would ANYONE use a partly working solution like MSAA.

    SMAA(FXAA) for performance.

    SMAA 2T (TXAA) for temporal smoothing and getting rid of that annoying pixel crawling.
    (I am guessing SMAA is sharper, but TXAA does better against crawling?)

    SMAA High, as the High-End solution
     

Share This Page