+ Reply to Thread
Page 2 of 3 FirstFirst 1 2 3 LastLast
Results 11 to 20 of 26

Thread: question about framerate and refresh rate

  1. #11
    Senior Member Flanger is on a distinguished road Flanger's Avatar
    Join Date
    Aug 2010
    Posts
    281
    I think that in any MPlayer FPS game vsync should be turned off. My LCD monitor can only support up to 75Hz. This means If I switch vsync on, I won't get more FPS than 75. It won't have tearing effect any more, but can you imagine playing quake on 75FPS? Besides as Lorfa has already mentioned, on top of that, vsync introduces some kind of input lag. I literally can feel that crosshair does not follow mouse as soon as it has to. I think even if I had 140Hz CRT monitor it would be the same.

    So I'd just turn off and forget about vsync and try to have as high physical refresh rate as possible. On my old LCD, I play with native resolution 1680x1050 and it has only 60Hz refresh rate (can't stand smudgyness of lower resolutions so have to sacrifice 15Hz), nevertheless I don't even notice tearing at all. Only if deliberately try to see it

  2. #12
    Senior Member mi will become famous soon enough
    Join Date
    Aug 2010
    Posts
    412
    Quote Originally Posted by julios View Post
    It's a shame qlive has capped the fps at 125, although I understand there may be physics dependency issues at framerates above 125.
    Yes. What a shame.
    Physics was framerate dependent in q3 - higher jumping at 333fps, weird effects at 1000, 2000 and 4000.
    But in QL the physics seems framerate independent - check jumping height/distance at different com_maxfps. So it would be much better to make the cap 250, 500 or even 1000.

    Quote Originally Posted by julios View Post
    I've heard anecdotes about capping framerate to refresh-1 fps, but not sure if this actually helps (or is possible in qlive).
    I tried refresh rate that matches QL 125fps-125Hz (with vsync off). It didn't work well, instead of tearing there was one horizontal line somewhere on the screen dividing two frames. It looked as bad as tearing, only the parts above/below the line looked smooth like with vsync on.


    One more tearing test you can try is
    devmap a map
    timedemo 1
    Then framerate is uncapped - it will show 125 in QL's counter, but use external osd like AfterBurner and then it shows 500 and more fps.
    .

  3. #13
    Senior Member Lorfa is a jewel in the rough Lorfa is a jewel in the rough Lorfa is a jewel in the rough Lorfa's Avatar
    Join Date
    Aug 2010
    Location
    Kepler-22b
    Posts
    8,369
    Quote Originally Posted by julios View Post
    Yes, have tested, and yes there is tearing. Although, I really only noticed this tearing using the health orb test, and it's not severe when actually playing. And yes, there is a significant difference with and without vsync, easily testable since you only have to toggle one command (r_swapinterval) and don't even need a vid_restart.
    Ok.. I might do a vid_restart anyways for good measure.

    Do the test I suggested. And do it with vsync enabled. I'm curious to see whether you notice a difference.
    I've run these tests before actually, and I don't see any difference in QL. I made sure it didn't say to force vsync off in my card drivers, and I ran every test I could think of. I actually came to the conclusion that it just didn't work on my system, or perhaps the vsync code was removed from QL.

    It might not be the case, but I mean, if I can't use vsync I'm not really worried about it.

    The bulk of the tests I ran on this were in quakeworld, and to say I ran a lot of tests is an understatement. Since the modern qw clients had independent physics, I was trying to find the best value of maxfps to use in combination with refresh rate to eliminate tearing, ideally give me a multiple of the sv_fps of 77, and also feature no noticeable tearing.

    In that situation I certainly noticed no tearing with vsync, but of course I didn't want to use it due to its latency. What I did was on dm3 go to the YA room and look at the vertical light beams where I would strafe back and forth. If there was any tearing it would show up there for sure.

    I would enter powerstrip, pencil in some values, go back in, adjust maxfps, rinse and repeat. I was at least able to move the tearing line to somewhere that I didn't really notice it, or its effect was very minor. I spent so many hours doing this that if someone plays my life experience back on a video tape and is fast forwarding, it's going to take a long time to get past that part.

    I settled on 153.918 Hz with a maxfps value and horizontal refresh value I don't remember.

    I was overjoyed that when I started playing QL, the tearing was not noticeable at the 153.9xx value, displayrefresh 150/800x600. I was disappointed when running 120 or 125 at 1024x768 resulted in tons of tearing, but 800x600 was still there for me and worked.

    I really hate tearing, I mean I just can't stand it. It's one of those things that makes me feel like hardware/software manufacturers just don't give a rat's patootie about people who want a low latency, tearing free gaming experience. I mean how long has it been known that digital video output with no motion blur features tearing if the fps/refresh rate aren't just so, and that vsync incurs a latency. We still have to run tweaks and tests to solve the problem, and deal with the thick skulled people who say things like "ur eye cant see moar than 30 fps" or "is only 25 ms lag that is like a 40th of a second man ur wastin ur time".

    I've heard anecdotes about capping framerate to refresh-1 fps, but not sure if this actually helps (or is possible in qlive).
    It isn't possible due to the 1000/x business.

    r_displayrefresh of course. Why would I use anything else when quakelive allows me to set it to what I want?
    r_displayRefresh is temperamental for me. It also appears to take only 1000/x values, and if you put in a value it doesn't like, it goes back to 60 Hz.

    Try to make sure that the refresh rate is what's set, like if there's a setting in your OSD to check it.

    With powerstrip though I can usually get any refresh I want, it just has to be near an accepted value of r_displayRefresh. This is why I'm getting 153-154 Hz instead of 150. The display OSD could be wrong I suppose, but it's also corroborated by powerstrip since I use the same values for the desktop. When I enter quake there is no "pop" which tells me there is no change in refresh rate or resolution. Although very small changes, such as 1-2 Hz can be made without "popping".

    fw900
    Awesome.

    I may experiment with lower res/higher refresh to see if that helps the tearing "problem".
    Ok, let me know how it works out.

  4. #14
    Senior Member Lorfa is a jewel in the rough Lorfa is a jewel in the rough Lorfa is a jewel in the rough Lorfa's Avatar
    Join Date
    Aug 2010
    Location
    Kepler-22b
    Posts
    8,369
    Quote Originally Posted by mi View Post
    Yes. What a shame.
    Physics was framerate dependent in q3 - higher jumping at 333fps, weird effects at 1000, 2000 and 4000.
    But in QL the physics seems framerate independent - check jumping height/distance at different com_maxfps. So it would be much better to make the cap 250, 500 or even 1000.
    It's not independent physics really per se. It's some kind of hack that allows the physics to remain constant up to 125 fps. They would definitely have to "retool" to get it to go higher without nasty side effects. Still I think it's worth it. Even the 150 fps I get in wolfcam is noticeable and desirable to me.

    One more tearing test you can try is
    devmap a map
    timedemo 1
    Then framerate is uncapped - it will show 125 in QL's counter, but use external osd like AfterBurner and then it shows 500 and more fps.
    Woah is this true? I'm going to try it.

    EDIT: OMG it does work!! That's so cool!!
    Last edited by Lorfa; 04-17-2013 at 06:27 AM.

  5. #15
    Senior Member pikaluva13 will become famous soon enough pikaluva13's Avatar
    Join Date
    Aug 2010
    Location
    East Lansing, MI
    Posts
    6,206
    Quote Originally Posted by Flanger View Post
    but can you imagine playing quake on 75FPS?
    I play with 60fps with VSync on, so I'd love to get 75 consistently.

  6. #16
    Senior Member quiet_L is on a distinguished road
    Join Date
    Dec 2012
    Posts
    136
    I enabled vsync last night. I consistently get 125fps and have a 120hz monitor (BenQ XL2420T)

    There is definitely mouse lag. Mouse movements feel way off.

    (Something to note: FPS kept alternating between what seems to be 117, and 125 with vsync on. I said seems because it skips back and forth so fast I can't even read it. vsync off, this doesn't happen!)

  7. #17
    Senior Member pikaluva13 will become famous soon enough pikaluva13's Avatar
    Join Date
    Aug 2010
    Location
    East Lansing, MI
    Posts
    6,206
    Quote Originally Posted by quiet_L View Post
    I enabled vsync last night. I consistently get 125fps and have a 120hz monitor (BenQ XL2420T)

    There is definitely mouse lag. Mouse movements feel way off.

    (Something to note: FPS kept alternating between what seems to be 117, and 125 with vsync on. I said seems because it skips back and forth so fast I can't even read it. vsync off, this doesn't happen!)
    I've never noticed mouse latency with vsync on, but I also use a wireless mouse that supposedly should have input lag as well...

    The fps alternating is probably normal and is probably stable at 125. I'm not sure why it sometimes does that though.

  8. #18
    Senior Member julios is on a distinguished road
    Join Date
    Dec 2011
    Posts
    167
    Quote Originally Posted by Lorfa View Post
    I've run these tests before actually, and I don't see any difference in QL. I made sure it didn't say to force vsync off in my card drivers, and I ran every test I could think of. I actually came to the conclusion that it just didn't work on my system, or perhaps the vsync code was removed from QL.
    I found the best way to verify whether vsync was enabled was by changing my refresh rate to 60, setting com_maxfps to 125, and having cg_drawfps 1. Toggling r_swapinterval immediately changed my framerate, as displayed in drawfps. Of course, I also set my nvidia display properties to allow the 3d application to control vsync.

    Quote Originally Posted by Lorfa View Post
    What I did was on dm3 go to the YA room and look at the vertical light beams where I would strafe back and forth. If there was any tearing it would show up there for sure.
    very cool.

    Quote Originally Posted by Lorfa View Post
    I settled on 153.918 Hz with a maxfps value and horizontal refresh value I don't remember.
    It seems like you're saying you set the vertical and horizontal refresh rate independently. This doesn't make sense to me, as the horizontal refresh rate is a function of the full vertical refresh rate and the current resolution (at least this is my understanding).

    Quote Originally Posted by Lorfa View Post
    I really hate tearing, I mean I just can't stand it. It's one of those things that makes me feel like hardware/software manufacturers just don't give a rat's patootie about people who want a low latency, tearing free gaming experience. I mean how long has it been known that digital video output with no motion blur features tearing if the fps/refresh rate aren't just so, and that vsync incurs a latency. We still have to run tweaks and tests to solve the problem, and deal with the thick skulled people who say things like "ur eye cant see moar than 30 fps" or "is only 25 ms lag that is like a 40th of a second man ur wastin ur time".
    Totally with you here. Not many people in the gen population who have interacted with a crt, let alone played a game like quake at a high level.

    Quote Originally Posted by Lorfa View Post
    r_displayRefresh is temperamental for me. It also appears to take only 1000/x values, and if you put in a value it doesn't like, it goes back to 60 Hz.
    If I have time, I'll do some experimenting this evening with powerstrip.

    Quote Originally Posted by Lorfa View Post
    Try to make sure that the refresh rate is what's set, like if there's a setting in your OSD to check it.
    Yea I only trust my OSD. Sometimes even my display properties report a refresh rate that is different from what OSD reports. For example, in 2304x1440, display props report 80 hz, but osd reports 75. And when I try to set it to 75 in display props, it gives me a panned display at 75. I think this might be at least partially due to the fact that I'm using a BNC connection (RGBHV), which doesn't support EDID. I may experiment with the regular DVI-VGA cable also.

  9. #19
    Senior Member Lorfa is a jewel in the rough Lorfa is a jewel in the rough Lorfa is a jewel in the rough Lorfa's Avatar
    Join Date
    Aug 2010
    Location
    Kepler-22b
    Posts
    8,369
    Quote Originally Posted by julios View Post
    I found the best way to verify whether vsync was enabled was by changing my refresh rate to 60, setting com_maxfps to 125, and having cg_drawfps 1. Toggling r_swapinterval immediately changed my framerate, as displayed in drawfps. Of course, I also set my nvidia display properties to allow the 3d application to control vsync.
    Ahh ok. It does the same for me so it does work. The difference between the two is barely perceptible at my 125 fps/153.9xx Hz (I consider myself very lucky here!).

    It seems like you're saying you set the vertical and horizontal refresh rate independently. This doesn't make sense to me, as the horizontal refresh rate is a function of the full vertical refresh rate and the current resolution (at least this is my understanding).
    With powerstrip you can do much! :-)

    You can't change the horizontal too much without affecting the vertical though.

    Yea I only trust my OSD. Sometimes even my display properties report a refresh rate that is different from what OSD reports. For example, in 2304x1440, display props report 80 hz, but osd reports 75. And when I try to set it to 75 in display props, it gives me a panned display at 75. I think this might be at least partially due to the fact that I'm using a BNC connection (RGBHV), which doesn't support EDID. I may experiment with the regular DVI-VGA cable also.
    Welcome to hell.

  10. #20
    Senior Member julios is on a distinguished road
    Join Date
    Dec 2011
    Posts
    167
    Quote Originally Posted by Lorfa View Post
    Welcome to hell.
    hah.

    Wow, that's amazing that you have found a tear-free refresh rate. Was it really a sweet spot deal, or just the highest one that your liyama could support?

    using r_mode -1, I managed to get 160hz going (1105x690), but it didn't really seem to help with the tearing. Gonna have to go the powerstrip route. Should I worry about experimenting with sync width, front porch, etc? And is powerstrip free these days?

+ Reply to Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts