Crysis the original and shocking FPS

Discussion in 'Videocards - AMD Radeon Drivers Section' started by JnS, Apr 22, 2011.

  1. Russ369

    Russ369 Master Guru

    Messages:
    621
    Likes Received:
    0
    GPU:
    EVGA GTX 1070 FE
    Crysis doesn't use physx

    fyi, we practically have the same system, im getting lag too... im thinking it might be the cpu bottleneck tbh
     
  2. Pill Monster

    Pill Monster Banned

    Messages:
    25,214
    Likes Received:
    9
    GPU:
    7950 Vapor-X 1100/1500
    ^How many 4Ghz CPU's do you think were around in 2007 when Crysis was released? Not many, if any.

    Fwiw I played Crysis with a 2.1Ghz X2 3800 & 6800GT @ 1600x1200, and still averaged 30fps.

    My .cfg file was fairly tweaked of course...
     
    Last edited: Apr 23, 2011
  3. Death_Lord

    Death_Lord Master Guru

    Messages:
    721
    Likes Received:
    8
    GPU:
    Aorus 1080 GTX 8GB
    Since Ati doesn't have 16x FSAA when you select 16x AA in Crysis 1 the Catalyst assigns 8x Edge-Detect AA to the game which turns into 24x samples, and because of that the Performance drops, 8X on crysis is the top, you will not get rid of the horrid unsampled transparencies anyway like the ones on trees for example.

    Crysis Warhead had a more agresive LOD that made the game look like if things were boiling as you come closer to them. :p
     
  4. The Chubu

    The Chubu Ancient Guru

    Messages:
    2,540
    Likes Received:
    0
    GPU:
    MSi GTX560 TwinFrozrII OC
    Whoa, 8x Edge Dettect + ingame antialasing could kill the perf all right.
     

  5. BlackZero

    BlackZero Ancient Guru

    Messages:
    8,878
    Likes Received:
    479
    GPU:
    RX Vega
    It's definately not a cpu bottleneck, lol. Sounds like crossfire is simply not kicking in or some other conflict somewhere.

    Here's my Crysis average fps using a Q6600 @3.6Ghz and a single HD 6950 unlocked to HD 6970 specs. Using the same ingame settings as the OP.



    23/04/2011 15:43:22 - Vista 64
    Beginning Run #1 on Map-island, Demo-benchmark_gpu
    DX10 1920x1080, AA=16x, Vsync=Disabled, 32 bit test, FullScreen
    Demo Loops=2, Time Of Day= 9
    Global Game Quality: VeryHigh
    ==============================================================
    TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
    !TimeDemo Run 0 Finished.
    Play Time: 54.34s, Average FPS: 36.81
    Min FPS: 25.65 at frame 1956, Max FPS: 46.06 at frame 987
    Average Tri/Sec: -10643915, Tri/Frame: -289183
    Recorded/Played Tris ratio: -3.17
    !TimeDemo Run 1 Finished.
    Play Time: 50.62s, Average FPS: 39.51
    Min FPS: 25.65 at frame 1956, Max FPS: 46.26 at frame 886
    Average Tri/Sec: -10874938, Tri/Frame: -275243
    Recorded/Played Tris ratio: -3.33
    TimeDemo Play Ended, (2 Runs Performed)
    ==============================================================

    Completed All Tests

    <><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

    23/04/2011 15:43:22 - Vista 64

    Run #1- DX10 1920x1080 AA=16x, 32 bit test, Quality: VeryHigh ~~ Last Average FPS: 39.51
     
  6. vf

    vf Ancient Guru

    Messages:
    1,868
    Likes Received:
    103
    GPU:
    AMD Radeon™
    Add that to the shortcut. "X:\xxxxxxxx\Crysis\Bin64\Crysis.exe -dx9"
     
  7. Ti3Kob

    Ti3Kob Master Guru

    Messages:
    682
    Likes Received:
    1
    GPU:
    980Ti @1505/7600
    I don't say he's bottlenecked. I say you're wrong saying CPU bottleneck do not exists. And sorry but I know perfectly what is a bottleneck and if you don't know what it is, just use a dictionary mate. I know it's full of words but it's a useful thing to have when you want to use a word you don't know.

    I was playing Crysis at launch with a single core 3700+@3GHz and a 7800 GTX, then a 8800 Ultra. Got almost the same performances on both cards, at tweaked medium settings (25 FPS avg on the 7800, and 30 on the 8800), yet stabler on the 8800 (pretty normal). Then I did own a Q6600, and even without any overclock, when the 7800 did not gain any perf boost, the 8800 got 30 fps more on average, +15/20 fps more with the CPU @ 3.8GHz. Then I was able to play on high settings, then tweaked very high settings.
    Sure, CPU bottleneck do not exist :nerd:.. You're wrong, sorry to say that.
     
  8. Jonup

    Jonup Active Member

    Messages:
    69
    Likes Received:
    3
    GPU:
    MSI GTS250 828/1944/2500
    Pill Monster, with all your due respect, if you look at the FC2 bench, the P2x6 1100T posts the same fps in all resolutions. It is definitely CPU bottleneck or some other bottleneck but it definitely it is not GPU bound. P2x4 observes similar phenomenon.
    That said I am not saying the JnS's problem is CPU bottleneck. I don't even like Crysis => I don't play it.
     
    Last edited: Apr 25, 2011
  9. GrandMax

    GrandMax Member Guru

    Messages:
    191
    Likes Received:
    0
    GPU:
    GTX 680 @ 1270Hz
    CPU Bottleneck hypothesis seems very probable.

    - CPU is not overclocked
    - My 1055t bottlenecks my pair of 6870.

    The easiest way to confirm my CPU bottleneck hypothesis is to check GPU usage with MSI Afterburner. If the graphic cards are maxed out, then we know it was false and that AA was too aggressive. If the cards are like at 50% or so, then we know they are limited by the CPU and AA can be excluded. Well, actually, AA can already be excluded since the OP removed it with little effect.

    In my case, i get about the same fps as the OP with my 1055t at 3.5GHz. One or two GPU doesn't change anything. Increasing AA and rez cranks the GPU usage but I can't quite reach more than 60%.

    r_MultiGPU does not change anything for me. The game detects by itself my setup. Just check with r_displayinfo and it should say multigpu already in the header.
     
  10. Pill Monster

    Pill Monster Banned

    Messages:
    25,214
    Likes Received:
    9
    GPU:
    7950 Vapor-X 1100/1500
    I never said they don't exist I said they're mostly a myth - in other words there is a tendency to automatically blame low fps on a bottleneck, when often this is not the case...it happens constantly on this forum...

    Even when bottlenecking is present it doesn't reduce performance but prohibits max performance..slight difference.

    @Jonup...wouldn't lowering the res be the easiest way to find out...? Just a thought..

    Remember Guru3D (I think or maybe LedHed linked to it) did a test about 2 years ago with Intel CPU at 4-5 different resolutions - the end result was about 2 fps difference across the board...
     
    Last edited: Apr 25, 2011

  11. Black_ice_Spain

    Black_ice_Spain Ancient Guru

    Messages:
    4,581
    Likes Received:
    14
    GPU:
    5700XT
    crysis = 2 core CPU.

    Best thing for it would be an i7/i5/i3
     
  12. BlackZero

    BlackZero Ancient Guru

    Messages:
    8,878
    Likes Received:
    479
    GPU:
    RX Vega
    I agree with Pill Monster, the 1090t is a modern processor, just because a certain game/application (such as FC2) gets better fps with a higher clocked/or intel processor doesn't mean there's a bottleneck. It just means the given application/game favours the other processor.

    That's like saying that, in certain guru3d reviews where benchmarks get higher fps for nvidia cards in comparison to AMD cards, the AMD cards are bottlenecked... yeah if you were to overclock the AMD card you would get higher fps but it's certainly not a bottleneck as the AMD card may get higher comparative fps in different benchmarks.

    A real bottleneck would be where a certain piece of hardware gets consistently lower fps in *most* situations and makes those applications/games unplayable.

    In this case, crysis, one of the most gpu intensive games available, is unlikely to get bottlenecked at 1920x1080 16xaa with a pair of 6970's and a 1090t. Check the review below of crysis warhead at max setting on a hd6990, even if a 1090t performs lower than the cpu used in that review, it's definately not going to be the limiting factor.

    http://www.guru3d.com/article/his-radeon-hd-6990-crossfire-review/16

    If we were talking about a pair of 6990's or gtx 590's then yeah it would have been quite possible that performance was less than you'd get with a better cpu but it definately wouldn't make the game unplayable. It would just mean you spent a truck load of money to play at a resolution that doesn't need that much gpu power.
     
  13. Krogtheclown

    Krogtheclown Master Guru

    Messages:
    960
    Likes Received:
    0
    GPU:
    FuryXfire 3440x14freesync
    Awh **** another CPU bottleneckn Quote! When will this bull**** stop!
    This is a GPU usage game get your **** straight!

    By the way make sure you launch this game from 32 bit folder 64 plays badly for my tri 6850 but 32 runs its 3 time faster.
     
  14. GrandMax

    GrandMax Member Guru

    Messages:
    191
    Likes Received:
    0
    GPU:
    GTX 680 @ 1270Hz
    Fps will always be limited by something. This something is referred to as the bottleneck, i.e. where the performance chokes and it will vary from game to game. In modern games, we are used to GPU bottleneck with high resolution and AA as it is most often the fps limiting factor.

    Crysis is unfortunately not using those extra processor of the OP's 1090t and the performance by thread is not as high as a i7, for example, especially at stock clocks.

    We don't have to debate this as the OP can quickly check and confirm. In my case, I can confirm the CPU (1055t at 3.5) is preventing my system to achieve higher fps. That is a real bottleneck Phil. I don't know how you can argue otherwise.
     
  15. Pill Monster

    Pill Monster Banned

    Messages:
    25,214
    Likes Received:
    9
    GPU:
    7950 Vapor-X 1100/1500
    If what you are saying was true, then a new quad core 1055T would need to have less performance per core than a dual-core CPU manufactured in 2006!

    Why not go back to Skt 939 then, if that'll improve performance?

    Come on...lets be realistic here shall we - I'm sure you don't really think that.....:3eyes:
     

  16. GrandMax

    GrandMax Member Guru

    Messages:
    191
    Likes Received:
    0
    GPU:
    GTX 680 @ 1270Hz
    Pill,

    The 1055t is also a 6 core. 45nm architecture.

    As far as I can tell, no one was playing Crysis DX10 very high at 1080p with AA back in the day.

    My fps ranges from 25-75fps but it is too often around 30. Increasing resolution and adding AA helps increasing GPU usage but not fps. I have heard multiple times people saying that AA does not do a performance hit in Crysis with very high end GPU setups and that is the reason.

    Many games work well with my Hex, but Crysis isn't one of them. Multi-threading is getting better, especially with DX11 so hopefully I will manage to use all six cores in BF3 and other major release this year. If it isn't the case, I'll go for a 2600k.

    Those are facts, it sucks but it is like that. The OP can reach decent fps if he cranks the clocks of his CPU, even on air.

    GrandMax
     
  17. BlackZero

    BlackZero Ancient Guru

    Messages:
    8,878
    Likes Received:
    479
    GPU:
    RX Vega
    GrandMax,

    Max fps in Crysis is not achieved with higher cpu clocks, it's a gpu limited game, adding AA/IQ will increase the load on the GPU and reduce FPS not increase it. If it was that CPU limited then reducing image quality would not result in increased FPS.

    Unless something has gone horribly wrong with the AMD hexa-core processors I can't see why you wouldn't see better fps at 1920x1080 with more gpu power.

    I even went and tested by reducing the game qaulity to medium just to see if there was a cpu limitation of some sort but the game acted as expected; the fps increased. I went from averaging between 30 and 40 fps at 1920x1200 very high with AA to 50-60 at medium. I'm not even using an i7, just a core 2 quad.

    However, I did notice a reasonable drop in gpu usage which has left some questions unanswered.


    Edit:

    Found this article

    http://www.techspot.com/article/73-crysis-performance/page7.html

    Seems overclocking the cpu core has very little effect on fps.


    and here

    http://www.guru3d.com/article/cpu-scaling-in-games-with-quad-core-processors/9


    At higher resolutions the cpu will have negligable effect, though I would have to agree that we have much faster newer processors and these articles are quite old, so again nothing conclusive.
     
    Last edited: Apr 26, 2011
  18. Pill Monster

    Pill Monster Banned

    Messages:
    25,214
    Likes Received:
    9
    GPU:
    7950 Vapor-X 1100/1500
    Mate look at what you're saying - your argument has lost all sense of logic. :bang:

    Wat does DX10 or DX11 have to do with a CPU bottleneck?
    How does a hexacore or quad CPU become bottleneck simply because it has 6 cores?
    How do you expect to see a fps gain by increasing the resolution and AA???
    And at 1920x1080 the CPU doesn't even come into the picture.


    And ur ignoring the most basic obvious logic - which BlackZero has just pointed out - seems he is the only person who understands where I'm coming from...kudos to him then, and well stated - better than I could have said it at least.. :p

    The bottom line is that you think a 6yr old dual core CPU is going to outperform a brand new hexacore simply because it has only 2 cores.
    By the same token you think a hex is bottlenecking the GPU because it has 6 cores....

    No disrespect - you may have convinced yourself with that story but you'll need to do better in order to convince me...I wasn't born yesterday my friend..;)

    Seems there's a lot of misunderstanding about what defines a bottleneck , and also how to spot one...




    EDIT:
    The OP already stated that his fps increased when AA was reduced, which proves the CPU is not holding back the GPU.
     
    Last edited: Apr 26, 2011
  19. Krogtheclown

    Krogtheclown Master Guru

    Messages:
    960
    Likes Received:
    0
    GPU:
    FuryXfire 3440x14freesync
    THe only time the word bottleneck should be used is when FPS come low enough to cause a problem. 100fps vs 300fps is considered a bottleneck by some here yet how in the **** does it matter at all in any game?
     
  20. flow

    flow Maha Guru

    Messages:
    1,011
    Likes Received:
    14
    GPU:
    Asus TUF RTX3080 oc
    A faster cpu will do better than a slower cpu, same goes for graphic cards, combine them and you can get a bottleneck. However, you still get the performance you payed for. Want more? Pay more. In that sense a bottleneck is just a word. Your cpu wont do worse because your gpu is slow, your gpu wont do worse because your cpu is slow, however, max graphics and still stuttering is not enjoyable.
    As for crysis, a beautifull game, is perhaps not that optimized as it should be. A very demanding game, still looking very good at maximum settings, still making things tough for highend machines. I suppose it's time not to use this game anymore as a reference for a new system. It probably never will do better at new tech.
     

Share This Page