|
On November 08 2011 06:26 Kr1pos wrote:
Whoever thinks enabling vsync (when not in conjunction with triple buffering) doesn't have a tangible impact on input lag must never have tried it in a twitched-based game like Quake or CS. Q3 with vsync is to me unplayable. It's probably not nearly as much of an issue in RTSes though, so I can see the argument for it there -- though I've never compared it myself.
Exactly. Screen tearing is not a big deal compared to input lag. But most people here run LCD's @60hz have mice running at 125hz, and dont have amazing frame rate. So that extra latency doesn't feel like much to them. But broodwar, or sc2 with 140 hz on a CRT with no input lag and a mouse clocked at 1000hz
|
Ok do whatever pleases you. Try it out and if the 'feeling' is off then just do what feels better.
Anyway all i wanted to say is there is no reason to tie the analysis of user input to the rendering of the display. They might have done it in old games (You're talking about Quake1 engine...) for reasons that are not valid today. And even if you do it the input lag is tiny.
For me all this stuff is just big number voodoo (1000Hz mouse, CS running at couple hundred FPS) to sell hardware, nothing else.
|
On November 08 2011 07:02 pecore wrote:
Ok do whatever pleases you. Try it out and if the 'feeling' is off then just do what feels better.
Anyway all i wanted to say is there is no reason to tie the analysis of user input to the rendering of the display. They might have done it in old games (You're talking about Quake1 engine...) for reasons that are not valid today. And even if you do it the input lag is tiny.
For me all this stuff is just big number voodoo (1000Hz mouse, CS running at couple hundred FPS) to sell hardware, nothing else.
Mouse polling rate is a very big deal actually. Pretty close to the number 1 aspect to a mouse in fact.
|
On November 08 2011 07:02 pecore wrote:
Anyway all i wanted to say is there is no reason to tie the analysis of user input to the rendering of the display. They might have done it in old games (You're talking about Quake1 engine...) for reasons that are not valid today. And even if you do it the input lag is tiny.
How do you propose one does that? Input lag is the delay between the sampling of the input until you see the changes on the display, so they are very much related. The result of the input (say, how far the mouse moved) will be the same either way and is of course unrelated.
And this isn't about engine implementation but the inherent implications of enabling vsync. Here's a comparison of the input lag in TF2 with and without vsync enabled (51ms average without compared to 79ms with on a LCD, 43ms to 84ms on a CRT. They're also testing Fallout 3, but there the input lag just plain sucks either way.).
For anyone interested in the subject this is another great article from Anandtech on triple buffering which when enabled gives you vsync without the input lag. Too bad it's seems to be primarily OpenGL-only for the moment. (Loving having it enabled in Rage though.)
|
On November 08 2011 07:18 Kr1pos wrote:Show nested quote +On November 08 2011 07:02 pecore wrote:
Anyway all i wanted to say is there is no reason to tie the analysis of user input to the rendering of the display. They might have done it in old games (You're talking about Quake1 engine...) for reasons that are not valid today. And even if you do it the input lag is tiny. How do you propose one does that? Input lag is the delay between the sampling of the input until you see the changes on the display, so they are very much related. The result of the input (say, how far the mouse moved) will be the same either way and is of course unrelated. And this isn't about engine implementation but the inherent implications of enabling vsync. Here's a comparison of the input lag in TF2 with and without vsync enabled (51ms average without compared to 79ms with on a LCD, 43ms to 84ms on a CRT. They're also testing Fallout 3, but there the input lag just plain sucks either way.).
Ok maybe my understanding of input lag is off. I viewed input lag as the latency between user input and when the input can be evaluated by the application which would effect the game state. And of course you could calculate that unrelated to the display. The display is always 'behind' the game state or an 'estimation' of the game state. This becomes even more complicated with multiplayer games over a network.
Anyway I see now what you mean. The link seems interesting I will check that out.
Although the results agree with me imho. You posted that the maximum measured difference in input lag was 40ms relating to the display. Nobody can 'feel' 40ms.. at least you'll have a hard time convincing methat you can. But this is not that important with feelings... if you're convinced then it's real for you and that is all that matters, so go ahead and do what feels best.
|
You'd be surprised how noticeable the difference is. Remember, we're talking full screen aim movement, not in-game (object) latency (where there is often also some latency-mitigating prediction going on to mask it). Anyway, sure, if you can't tell the difference feel free to use whatever. And as I said, it probably doesn't matter much in RTSes anyway. But if you ever feel like trying out Quake (quakelive.com), give it a try (r_swapInterval 1 in the console). ^^
|
I don't even know what it is called input lag. If anything OUTPUT lag makes more sense. Output delay is probably most optimal since lag is typically reserved for networking conditions.
|
I'm pretty sure that Quake 3* physics made you run/accelerate slightly faster with specific FPS settings, due to rounding errors. Something like 125 FPS made you, on average, a bit faster. It was actually affecting the game. There were other key settings, like 97 and 71 IIRC, which weren't as fast, but were faster than default, and way better than 60FPS (Vsync).
In SC2, there doesn't seem to be anything of the sort, so settings above 60 or 75FPS shouldn't change anything. However, I'm fairly confident that, at least in StarJewel, having over 80FPS is advantageous. 125 probably isn't any better than 90, but 80 is definitely better than 60.
With that in mind, I try to keep my FPS at least at 80, maybe 60 very, very lategame. I think it's pretty noticeable and annoying whenever FPS dip below 60.
*Vanilla, they might have changed it for Quake Live, but IIRC, it was still in at the time they released the source.
|
This is kind of offtopic but:
Does LCD or Plasma have a slower output (like time it takes for action to appear on screen from the pc or console) than a fat TV? I remember that you are supposed to use non LCD or Plasma (either both or one of them) for fighters.
Is this true?
|
On November 08 2011 08:13 Eppa! wrote:
This is kind of offtopic but:
Does LCD or Plasma have a slower output (like time it takes for action to appear on screen from the pc or console) than a fat TV? I remember that you are supposed to use non LCD or Plasma (either both or one of them) for fighters.
Is this true?
The response value for LCD monitors is often 2ms or sometimes even lower. 500 FPS is reality perfect since the human synapse isnt capable of processing information faster than that. Thus, a monitor with 2ms (1000/500 = 2) response should in theory make maximal use of the human eye. Perhaps that rule of thumb was written long ago, when response times of 25ms and so forth were common. Hertz being equal of course. Hertz matters too. Higher the better, and we have 120hz LCD now, once again though after about 60 the rules start to shift.
I could be wrong though.
|
On November 08 2011 07:02 pecore wrote:
Ok do whatever pleases you. Try it out and if the 'feeling' is off then just do what feels better.
Anyway all i wanted to say is there is no reason to tie the analysis of user input to the rendering of the display. They might have done it in old games (You're talking about Quake1 engine...) for reasons that are not valid today. And even if you do it the input lag is tiny.
For me all this stuff is just big number voodoo (1000Hz mouse, CS running at couple hundred FPS) to sell hardware, nothing else.
There is one very important reason to tie analysis of user input to render loop. And that is synchronization, there is no point of reading input if frame is slow to draw because then you would still have to give time for render to catch up. So since quake 1 times game logic and input is done between drawing frames. There is no benefit to have it work in parallel but causes huge problems.
|
I remember I started performing extremely well in CoD4 when I got an FPS config so that I could run at a constant 125 fps because that game ran on the quake 3 engine. My input was way more accurate and my ping was lowered too. I'm pretty sure the same would apply to SC2, higher fps = better input.
|
On November 08 2011 03:42 Cyro wrote:Show nested quote +On November 08 2011 03:21 ZiegFeld wrote:On November 08 2011 01:15 Souljah wrote:On November 08 2011 00:59 -Archangel- wrote:
Trust me you cannot feel the difference and the eye cannot catch more then 36fps (or something around that). Maybe you are talking about a refresh rate of your monitor. What you might be feeling is drops in fps on your computer (some computers can have big average fps but can have drops). Try to turn on vsync to avoid those. dumb. obviously you've never seen the game at 120hz. Aye, my m17x has a 120hz screen. Smooth as fuck. I bought an m17x like 2.5 years ago... they only had "i7" cpu's avalible which were 1.6 and 2ghz, it chokes out and cant maintain 60fps in 1v1s on lowest settings with sli gtx260m's. They even removed the sli option because of heat issues that cause the laptop to go from cold boot to thermal shutdown in 2-3 minutes or so unless you explicitly limit the framerate in every single game you play LOL sucks for you, my quad core CPU has never so much as hiccuped. Only downside is the graphics card that I haven't upgraded to 560m (460m atm), so I have to play on medium/low graphics to maintain 120fps in 200/200 battles. The R3 is damn near flawless.
|
I al reading through this and I must be missing something.
How is it advantageous to have more than 60FPS on a 60Hertz monitor?
I am genuinely curious because I currently use vertical sync specifically because my thought process was to lock the FPS at the refresh rate of the screen since my computer can easily put out a constant 60FPS no matter how crazy the battles get.
|
On November 08 2011 09:03 Bizarro252 wrote:
I al reading through this and I must be missing something.
How is it advantageous to have more than 60FPS on a 60Hertz monitor?
I am genuinely curious because I currently use vertical sync specifically because my thought process was to lock the FPS at the refresh rate of the screen since my computer can easily put out a constant 60FPS no matter how crazy the battles get.
It is not. However, if you had a CRT (which can go above 60hz) or a 3D-ready LCD (which works at 120hz), it would be advantageous.
In any case vsync should always be off since it will add some input lag.
Hope this helps 
|
On November 08 2011 09:03 Bizarro252 wrote:
I al reading through this and I must be missing something.
How is it advantageous to have more than 60FPS on a 60Hertz monitor?
I am genuinely curious because I currently use vertical sync specifically because my thought process was to lock the FPS at the refresh rate of the screen since my computer can easily put out a constant 60FPS no matter how crazy the battles get.
There might be a bit less input lag. Its worth a try, usually you will notice the difference, if there is any for SC2, right away if you are used to playing with the old settings.
I feel the whole thing isn't that critical for SC2. Since you're only moving a cursor and not turning the whole screen, 60fps@85Hz are sufficient for me. I tried withouth framecap @100Hz and it felt smoother, but I think it is not really worth it to torture my CRT for it.
CSS though I play at 160Hz and wouldnt accept a frame less, so unbelievably smooth and direct.
|
I al reading through this and I must be missing something.
How is it advantageous to have more than 60FPS on a 60Hertz monitor?
I am genuinely curious because I currently use vertical sync specifically because my thought process was to lock the FPS at the refresh rate of the screen since my computer can easily put out a constant 60FPS no matter how crazy the battles get.
I might be repeating myself a bit, but it depends a bit on what the 60 fps is. Usually the framerate is far from constant during the game (unless it's capped to a lower value than the hardware can manage). So if the avearage fps is 60 it might mean that you actually have situations where the framerate drops to a lot lower value. Time between frames is actually the value humans should be able to sense or feel (choppy performance) not the amount of frames per second.
So strictly speaking if your machine can maintain a constant framerate above 60 then there are probably no noticable benefits to having a higher framerate on a 60 hz monitor and VSync should work ok as well. According to my experience, Vsync sometimes becomes problematic at lower framerates so I like to usually turn it off because I've never seen any benefits of having it on (this might be a personal thing though and probably differs depending on the hardware). The refresh rate is more of a CRT issue actually as they are in some sense "blinking" all the time, so they start to flicker at lower refresh rates.
I kind of doubt 60 vs 120 Hz gives much of a difference in user experience. You need higher refresh rates for 3D shutter glasses though, as they effectively halve the frame rate. From what I've read the 120 Hz displays are mostly about recreating the 24 fps movie format more smoothly (the 24 fps doesn't fit into 60 hz evenly unfortunately).
|
On November 08 2011 09:51 Hundisilm wrote:
I kind of doubt 60 vs 120 Hz gives much of a difference in user experience. You need higher refresh rates for 3D shutter glasses though, as they effectively halve the frame rate. From what I've read the 120 Hz displays are mostly about recreating the 24 fps movie format more smoothly (the 24 fps doesn't fit into 60 hz evenly unfortunately).
On a CRT there is a huge difference, and I am not talking about the 60Hz flickering. There is already an easily noticeable difference between 85 and 100Hz in SC2. It is smoother, and the mouse feels a little bit slower, although it has the exact same sensitivity. I believe that is because you feel comfortable controlling higher sensitivity at higher monitor frequencies.
In CSS i would also undershoot all the time after switching from 85 to 120Hz, and then again after switching from 120Hz to 160Hz. I had to figure out my perfect sensititvity again (which takes quite a while of testing, if you are meticulous about it).
It would be a bad thing if 120Hz (or 200Hz) LCDs wont be able to recreate that smooth feeling. Good CRTs are getting hard to come by, and 60Hz LCDs were always choppy in comparison, at least for CSS and Quake.
|
No, it's not normal that the game is more fluid at 120 frame per second as your eye can only see 60 frame per second. I can see the difference from 40 to 50 or 60, I can't tell how many frames per second my eye see but I know when its not fully fluid, aka 60 fps, it's just something that I got used to.
|
On a CRT there is a huge difference, and I am not talking about the 60Hz flickering. There is already an easily noticeable difference between 85 and 100Hz in SC2. It is smoother, and the mouse feels a little bit slower, although it has the exact same sensitivity. I believe that is because you feel comfortable controlling higher sensitivity at higher monitor frequencies.
In CSS i would also undershoot all the time after switching from 85 to 120Hz, and then again after switching from 120Hz to 160Hz. I had to figure out my perfect sensititvity again (which takes quite a while of testing, if you are meticulous about it).
It would be a bad thing if 120Hz (or 200Hz) LCDs wont be able to recreate that smooth feeling. Good CRTs are getting hard to come by, and 60Hz LCDs were always choppy in comparison, at least for CSS and Quake.
Yes I meant that in a LCD context only ofcourse (the frequencies on LCD's and CRT's have rather different meanings I'd say). Some people are probably more sensitive about such things unfortunately (my condolances to you), but I believe a majority of people wouldn't really be noticably happier if the refresh rates were higher.
PS. I have very limited experience with higher refresh rates (opinion based more on differences between lower rates), but people seem quite happy with their higher rates generally.
PS.PS. I did some internet surfing and it seems that it is generally accepted that CRT>120Hz LCD>60 Hz LCD for FPS's. Sad the 120 Hz LCD's are quite expensives still though.
|
|
|
|
|
|
EPT
