NVIDIA (and possibly AMD soon) PURPOSELY ADDING INPUT LAG!!!!!!!

Discussion in 'General Discussion' started by Spinelli, May 25, 2013.

  1. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    32
    What? Your reaction time can be great, if there is a consistent and constant delay between all controller inputs and physics engine then it'll cause issues, and many people can notice this, and notice when it improves.
     
  2. liebestod

    liebestod Registered

    Joined:
    Jan 9, 2012
    Messages:
    142
    Likes Received:
    2
    You're incorrect if you think people can't tell the difference in even one frame of input lag. If you think this is the case, you have the entirety of the competitive arcade gaming community to answer to. You'll find plenty of people that will easily be able to double-blind test with high accuracy in distinguishing a lagless emulated game versus one with 1-3 frames of input lag. If you're arguing that people are "FOS" when they say they can distinguish this, sorry -- I don't think so.

    Also, what could be lost from making sure that a game is as responsive as possible? Why wouldn't you want things to respond as quickly and optimally as possible? We're all after the same thing here, when you think about it... a better and more tight-to-control sim.

    edit: I should add that 1 frame is going to be tough for some people to notice but when you get into the area of 3+ frames, it starts to get obvious.

    Exactly.
     
    Last edited by a moderator: May 27, 2013
  3. martymoose

    martymoose Registered

    Joined:
    Aug 10, 2012
    Messages:
    461
    Likes Received:
    13
    I'm not saying there is no difference at all (though its very minimal to me) between sync on and off. I prefer a rock solid 2560x1440 @60hz compared to higher rates and tearing. I dont have a 120hz screen and was between a good hi res pro monitor, High colour accuracy not as in pro gamer whatever that is lol. I run a fast setup with an overclocked Titan so frames arent an issue and I tried running it with consistent frames well over 300fps and compared to the way I have it setup with sync, the lag is not noticeable and it has no effect on my laptimes. I did go through lots of trial and error to get the sync to work as I liked and have made some changes in the plr file as default v-sync is pretty bad. My frame rate pretty much never goes below 60p no matter how many cars are on track all maxed out at 2560x1440 and thats how I like it. Frame rate fluctuations are also an issue and that's why guys obviously lock frame rate to 120 or whatever but Id rather get the best image my monitor can display even if thats at 16ms its fine for me smooth as butter.

    I hate screen tearing and micro-stutters and I sim race for fun and dont care if I am a few tenths faster using some settings that look terrible to me and take away from the immersion. Its good that we have choices and there is no right or wrong way of doing things so no need to try and force your views on others as if its the only way of doing things. If someone beats you with v-sync on though it means you are way slower so maybe go down to 480p min settings to help gain another couple of tenths, who cares if you cant see the next corner but it should be faster. :D
     
  4. Noel Hibbard

    Noel Hibbard Registered

    Joined:
    Oct 5, 2010
    Messages:
    2,744
    Likes Received:
    40
    I seriously don't understand why this topic is always so hard for people to understand. No one wants to comment on my video with VSync on vs off? VSync doesn't always cause a single frame of lag. In my video you can very clearly see there is probably a lag of about 100ms.

    Here is another analogy that you guys can totally ignore.

    Which scenario would you chose:

    A: someone shooting at you with a paint ball gun while your eyes are open.

    B: someone shooting at you while your eyes are shut and then opening them right before impact.

    It is about having a specific amount of time to make a correction before it is too late. Why deliver the warning late if we don't have to?

    Yes VSync looks nice. Yes running over 60fps isn't going to be seen on a 60hz monitor. Yes there are other forms of latency in a sim. Why not cut out some latency if possible. Tearing is very minimal if any at all if you cap the framerate at 60 or some multiple of that. I get zero tear and zero stutters in rF1. Stutters with VSync off are caused by the CPU maxing out as it tries to keep the physics engine in sync with the gfx. If your CPU is running 100% then any background task is going to take time from rF1 in cause a stutter. You can't simply turn off VSync and call it a day. You have to cap the rate another way. In rF1 you use "Max Framerate=-60" in rF2 you use "Max Framerate=60". You find this setting in the PLR file. Personally ISI needs to get this setting moved into the GUI. My i5-2500K goes from 80% when uncapped with VSync off and uncapped to 22% with VSync off but capped at 120.

    I should do another video for people to not watch of my tear free, lag free system with VSync off.

    It seams only the iRacing community understands lag. iRacing comes defaulted to having VSync off and a Framerate cap of 80fps. All adjustable within the GUI. They understand the benefits of not running VSync but also understand that there are tons of side effects to running without some sort of Framerate cap.

    But anyways, continue on debating about human reaction time. ;)
     
  5. martymoose

    martymoose Registered

    Joined:
    Aug 10, 2012
    Messages:
    461
    Likes Received:
    13
    Like I said I prefer it one way after having tried both, you like it your way and cant see why anyone would do it differently. It doesn't bother me if others do it my way or any other way we can all do what we prefer, it also doesn't matter how the software is set by default as long as there are options for all to be happy. I always tweak everything to my preferred methods and find what works best for me as I am sure you do too, there are more then one way to skin a cat.

    If your way makes you faster then that's fine with me I can drive fast enough for me and I can run consistent competitive lap-times so I have no reason to sacrifice image quality if I am happy with what I get. My monitor is still on with v-sync enabled so saying its like having your eyes closed isn't the best analogy I think.

    Id hate for any one way to be forced or any options to be removed, the more options the better and like the OP is really worried about this setting in drivers Id hope for those guys its possible to be disabled as it seems a huge deal for some.
     
  6. SPASKIS

    SPASKIS Registered

    Joined:
    Sep 7, 2011
    Messages:
    3,155
    Likes Received:
    1,426
    Everybody talks about gfx input lag. But what about FFB lag? For me it is much more important than graphics. If the graphic delay is constant you trend to anticipate the rotation of the wheel when you realize you are missing the appex. However, when you lose grip you need a quick countersteer to balance the car back again. This is a reflex reaction that goes directly from your arms to your backbone where reflexes are computed. Does anybody know how much ffb lag do we have? Are there differences between racing wheels (as in monitors)? If you have poor graphic framerate or excessive lag ffb has the same framerate and lag problems?

    Enviado desde mi GT-I9505 usando Tapatalk 2
     
  7. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    36
    Please look at my bf3stats history graph of kdr page: http://bf3stats.com/stats_pc/DrR1pperUk/history#kdr

    [​IMG]

    I would like to bring your attention to the sudden increase in my average (i.e. running) KDR that started towards the end of february, the exact time i bought a ASUS 2ms 120hz monitor with light boost (upgraded from a Hannspree 5.5ms 60hz monitor). Please ignore the wild peaks and troughs at the beginning of this graph, this was a period of first learning to play that game well and then the updates that came along to stop the weapons having as much power and accuracy (making it tougher to kill as many people in a single clip hence i died more). After the first 9 months of game updates the game finally smoothened out with fairly consistent weapons handling and predictability once again. This is represented by the consistently stable kdr throughout the long middle period of the graph.

    Now, back towards where i bought the ASUS 2ms 120hz display. My running average KDR had gone from a very stable 1.85 to a sudden and rapid increase to 1.98 (last recorded gameplay exactly a month ago). Now this may not sound a big jump on its own but you must realize that this is my RUNNING AVERAGE and so it becomes very difficult to make any changes to your running average kdr the more kills and deaths you rack up over time. Up till before my new monitor, my total kills had been 38,000 against 19,000 deaths. You can imagine to break free of this would requires a ton of effort, either by 1) focusing on trying not to die as much or by 2) carry on doing exactly as i played before (e.g. technique wise) but somehow managing to get many more kills per death.

    Well, i can empirically show that the latter has been made true by upgrading my monitor and lowering my input lag considerably.

    We know that my running average stabalised at around 1.85 which meant my non running average kdr per game was also in fact around 1.85. However, during those 3 months using the 2ms 120hz display I racked up an extra 7,376 kills to 2,745 deaths. The non-running average kdr per game during those 3 months was 2.69. Not only does this empiracally show the difference lower input lag makes but i consciously felt it whilst playing...my actions were quicker and as i have said in the past on other threads having this much lower input lag felt like time slows down. I know how pseudo-ee that sounds but i just have so much more time to think and react compared to before. If i go around a corner, i now get the jump on the guy almost 9 times out of 10. There's even enough time for me to see how it hasn't quite registered to him yet that i'm an enemy and i have a quick chuckle to myself before placing a round in this head. I honestly feel bad sometimes as i think it's such a huge advantage over others that it's almost like a form of cheating.

    Your can translate these performance increases directly to any game/simulator.

    Now, if you honestly don't care about this performance benefits, then that is ofc fine.
     
    Last edited by a moderator: May 27, 2013
  8. osella

    osella Registered

    Joined:
    Jan 11, 2012
    Messages:
    864
    Likes Received:
    26
    If you use realistic FOV you usually HAVE to use virtual mirrors because on 1 screen actual car mirrors just won't fit. We could debate infinitely who uses overall less realistic setup, in my point of view it's you if use FOV of 60degress etc etc.
    And headphones? How the hell make headphones anything less realistic..?
     
  9. zenrael

    zenrael Registered

    Joined:
    May 21, 2013
    Messages:
    61
    Likes Received:
    0
    It is much easier to visualise the problem if, like others, you consider it in distance rather than time.

    At 100kph, your car will travel 28cm in just 10ms. Meaning if you sight your turning point, and turn, your car has actually travelled a further 28cm before the turn is executed. You run (slightly) wide and likely miss the apex.

    Sure, you can learn to drive with this disadvantage by simply turning earlier. Instinctively after a few laps you know when to turn in etc. But the problem truely manifests when you try to control a slide...

    Lets say the back end swings out at pi rad/s (so it takes a full second to spin 180 degrees). Once you notice the car begin its slide, with a 10ms latency it's already spun 1.8 degrees. A skip barber is ~365cm in length; that 1.8 degrees equates to roughly 11cm of lateral travel - the back end has stepped out 11cm more than you think it has..

    Try telling me that is negligible. It isn't. And we're talking about 16.5ms here, not 10ms. Infact some input latencies that i've seen around these forums have been as high as 40-80 in extreme cases. At 80ms input latency, your lovely little skippy has stepped out just under 90cm at the back before you are even told it is sliding.

    I'm sure i could be wrong here, it's early (for me, anyway, haha) - but if i am do try and correct me.

    John
     
  10. Noel Hibbard

    Noel Hibbard Registered

    Joined:
    Oct 5, 2010
    Messages:
    2,744
    Likes Received:
    40
    My analogy is perfect. It is exactly like having your eyes closed. What is happening in realtime (physics) is ahead of what you see. Essentially like closing your eyes with the car sliding a bit before you even see it.
     
  11. Noel Hibbard

    Noel Hibbard Registered

    Joined:
    Oct 5, 2010
    Messages:
    2,744
    Likes Received:
    40
    Great post John. Finally someone with a grasp on the problem. Someone will pick apart your numbers though rather than just following your logic.
     
  12. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    36
    great post zenrael, really puts it into even greater perspective.
     
  13. deak1944

    deak1944 Guest

    Synchronize Frame="1.00000" // Extrapolate graphics using estimated render time in attempt to more accurately synchronize physics with graphics, 0.0 (off) - 1.0 (full)

    So what does this do?
     
  14. zenrael

    zenrael Registered

    Joined:
    May 21, 2013
    Messages:
    61
    Likes Received:
    0
    Interesting. I'd like to know this too.
     
  15. Lazza

    Lazza Registered

    Joined:
    Oct 5, 2010
    Messages:
    12,345
    Likes Received:
    6,572
    You would have to think this would somewhat negate the lag you have when your motion is fairly constant - so the turn-in 'lag' described above, and the visual lag when under constant rotation, would be reduced. How far the game might go in predicting non-constant motion is anyone's guess.

    On the other hand, if it's using only the frame time at maximum, it would seem it won't counteract the complete lag you get since the frame time is only a part of it.
     
  16. Axeslayer

    Axeslayer Registered

    Joined:
    Jan 14, 2012
    Messages:
    178
    Likes Received:
    4
    Still using an old CRT here - that's input lag ?:D
     
  17. Satangoss

    Satangoss Registered

    Joined:
    Jun 2, 2011
    Messages:
    1,123
    Likes Received:
    7
    He's just trolling. The fact you have lag from your eyes to your arms doesn't mean you'll want to have an additional lag in your GPU drawing too. This argument is just ridiculous. Ignore it.
     
    Last edited by a moderator: May 27, 2013
  18. zenrael

    zenrael Registered

    Joined:
    May 21, 2013
    Messages:
    61
    Likes Received:
    0
    Good for you! I do heartily miss my old 22" CRT.

    It was so old and worn out, I had to heat the back of it up with a hair dryer before turning it on so the contacts on the badly soldered joints would expand and conduct, lol.
     
  19. Nand Gate

    Nand Gate Banned

    Joined:
    Jan 21, 2012
    Messages:
    471
    Likes Received:
    12
    Not empirical. The argument is how MUCH lag can a given individual notice. The OP is about 1fps. He will not notice. Sorry to argue against the raging input lag mob. Yawn.
     
  20. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    32
    The 16.7 ms makes a difference, it did to me back in the day when I never really knew about input lag. I felt it going from a CRT to a 20ish ms input lag monitor. Then when I couldn't handle it any more and got a rated 3.3 ms input lag monitor (Samsung PX2370) I immediately felt the difference, it was so much closer to my CRT driving experience.
     

Share This Page