GTX 680 pre rendered frames 0?

Discussion in 'Videocards - NVIDIA GeForce Drivers Section' started by krojf, Apr 11, 2013.

  1. krojf

    krojf Guest

    Messages:
    2
    Likes Received:
    0
    GPU:
    corsair vengeance duo 8gb
    So, I tried to follow the guide: http://forums.guru3d.com/showthread.php?t=377158

    Downloaded some of the quadro 29x.xx driver's... Unpacked as described.
    Open the INF. file.

    Added "NVIDIA_DEV.1180 = "NVIDIA GeForce GTX 680" to the [Strings] -

    And.

    "%NVIDIA_DEV.1180% = Section002, PCI\VEN_10DE&DEV_1180" -

    to the "NTamd64.6.1" section as I have Win7 64bit, I close and save it.
    Run the setup.exe as administrator, though it fails and shows the "Failed to install etc" screen

    Am I doing something wrong here? Yep I'm a newbie to modding drivers.
     
  2. rewt

    rewt Guest

    Messages:
    1,279
    Likes Received:
    15
    GPU:
    Yes
    Try basing your mod on GeForce drivers rather than Quadro drivers. It's kind of important you choose the correct Section###
     
  3. Prophet

    Prophet Master Guru

    Messages:
    865
    Likes Received:
    34
    GPU:
    Msi 680
    The earliest I managed to install was 297.03. I prefer the 297.16 though.

    Inf: https://skydrive.live.com/?cid=96b6875a64fe5c5a&id=96B6875A64FE5C5A!118&authkey=!AAxoIZZ9_GMwwgs
     
  4. krojf

    krojf Guest

    Messages:
    2
    Likes Received:
    0
    GPU:
    corsair vengeance duo 8gb

  5. rewt

    rewt Guest

    Messages:
    1,279
    Likes Received:
    15
    GPU:
    Yes
    NVIDIA claimed over and over again that maximum pre-rendered frames 0 didn't make any sense. All I can figure if this is true, the increase in latency perceived @1 must be balancing the latency caused by the reduction in FPS @0.

    They've recently invested resources in FCAT, perhaps something similar will come for input lag analysis.
     
    Last edited: Apr 13, 2013
  6. Rapyer

    Rapyer Guest

    Messages:
    218
    Likes Received:
    2
    GPU:
    1080ti
    zero frames to rendeer ahead you get a lot of screen tearing on a slower card. leaving it at 1 is much better, plus more performance. zero has no mouse delay, but overall the fps suffers and it lookes really bad and feels bad

    i used to experiment a lot with my 8800 in oblivion to make sure it ran somewhat smooth, and zero is just pointless in nearly all games
     
  7. Prophet

    Prophet Master Guru

    Messages:
    865
    Likes Received:
    34
    GPU:
    Msi 680
    If the card is slower you get tearing no matter what.

    0 feels good on the games I have tried it on, and in online games (fps in my case) that require fast reflexes its defintely nicer than 1+.

    And the 8800 isnt exactly a recent card.
     
  8. Rapyer

    Rapyer Guest

    Messages:
    218
    Likes Received:
    2
    GPU:
    1080ti
    its 1 frame. your reflexes wont matter.

    you have the wrongest misconceptions about everything, everything you say here is placebo talk
     
  9. Iruwen

    Iruwen Guest

    Messages:
    307
    Likes Received:
    0
    GPU:
    EVGA GTX 670 FTW
    Legitimate ****storm in 3... 2...
     
  10. rewt

    rewt Guest

    Messages:
    1,279
    Likes Received:
    15
    GPU:
    Yes
    Apparently zero had no mouse delay, at least until NVIDIA took it away..

    I agree performance can suffer badly at zero (that's my theory why NVIDIA decided to take it away in the first place), but that ultimately depends on the system/game.
     

  11. Prophet

    Prophet Master Guru

    Messages:
    865
    Likes Received:
    34
    GPU:
    Msi 680
    It´s 10 ms @ 100 hz, so it matters and its easily noticeable.
     
  12. rewt

    rewt Guest

    Messages:
    1,279
    Likes Received:
    15
    GPU:
    Yes
    I can't imagine all NVIDIA's employees failing to experiment with input lag to find the truth, but that appears to be the case judging by what yourself and other gamers have said ever since the 0 setting was removed.

    It seems they need to hire this guy:

    I've followed similar methodology and there really is a clear difference between pre-rendered frames 0 and 1. In the above example, the console window would lag slightly less behind the mouse cursor @0 than @1 (it's most easily detected with vsync enabled).
     
    Last edited: Apr 16, 2013
  13. Prophet

    Prophet Master Guru

    Messages:
    865
    Likes Received:
    34
    GPU:
    Msi 680
    Url?
     
  14. Black_ice_Spain

    Black_ice_Spain Guest

    Messages:
    4,584
    Likes Received:
    17
    GPU:
    5700XT
    they should let users pick 0 if they want, after all its an (very)advanced tweak. If people wants to make errors... just let them.


    anyways in that test, you should record it (with an screen recorder like fraps?/mirillis action, i don't know if those will be able to show it or not, depending on which stage they record)

    and then overlay both videos (0 and 1) placebos can be risky. And btw, i've read you speaking about 10ms as the time added from 0 to 1, this could not be entirely true.

    Apart from that, it should be the way that the user in the video is asking in order to measure it.

    Probably by moving the pointer by some kind of software will be enough to do the same test twice.

    After all, when you set 0, you dont paint a frame in the time t=0, but when it finishes rendering, which could be t=0.7, compared to the t=1 which would be the next frame (so when it's painted after the buffer)
     
    Last edited: Apr 16, 2013
  15. Rapyer

    Rapyer Guest

    Messages:
    218
    Likes Received:
    2
    GPU:
    1080ti
    what you can do to reduce that input lag is force higher usb polling rate

    if your framerate is below monitor refresh rate, set it to 1 always. if it's more than that, you can set it to 0, but prepare for tearing and framerate up to 20% slower

    and if you really want to reduce it, prepare to fork money on a good display

    and if you use sli - never use 0

    http://www.anandtech.com/show/2803


    also it largely depends on games you play
     
    Last edited: Apr 16, 2013

  16. Prophet

    Prophet Master Guru

    Messages:
    865
    Likes Received:
    34
    GPU:
    Msi 680
    Most people who game use 500 or 1k hz on the usb. Thats a 1 ms difference. Although I do prefer 1k hz it comes at the cost of slightly less stable polling rate. Ie 500 is a bit less fast but smoother. Most good gamers use 500.

    I use 0 on games where I need to be fast, I dont have any tearing though. 150< fps @ 120hz. Do you have something to support your other claims about setting 0 or 1 depending on your fps in regards to your refresh rate?

    Actually even though you probably can set 0 in the settings you can probably not use 0 with sli.
     
  17. The General

    The General Guest

    Messages:
    3,108
    Likes Received:
    5
    GPU:
    FirePro W5000 2GB
    i mess around with that settings with my 550 ti and i saw no difference in performance whatsoever, crysis 2 for example: 1600x900 + hardcore settings dx9 30-80, same at default and at 0. so 3 is best i guess.
     
  18. Rapyer

    Rapyer Guest

    Messages:
    218
    Likes Received:
    2
    GPU:
    1080ti
    they removed the 0 render ahead in 300 series drivers, if you set it to 0, it defaults to default 3

    you need to use older drivers to set 0

    hence 1 is the best compromise i feel

    also quoting manuelg (nvidia rep here)


    I asked our engineers and their reasoning for removing "0" was because there was practically zero difference in setting this to either "0" or "1". The result was the same.

    Manuel
    NVIDIA Forums Technical Advisor
     
  19. Prophet

    Prophet Master Guru

    Messages:
    865
    Likes Received:
    34
    GPU:
    Msi 680
    Yea I know he claims that. 10 ms + is not 'practically zero' imho. I was wondering about the other stuff though, having 0 / 1 set if you have less fps than hz (iirc).

    It´s a question of money, not of giving users good service.
     
  20. Rapyer

    Rapyer Guest

    Messages:
    218
    Likes Received:
    2
    GPU:
    1080ti
    i posted a link to anandtech, have a read. it mentions it in conclusions
     

Share This Page