The performance of the Spring engine

[reivanen]

What ever you say, there is no way you could in any way feel or hear FPS rating higher then what your display is capable of updating.

So a typical TFT screen only updates 60 times a second, so 60fps is exactly as smooth as 125FPS. (Well, probably 60fps vsynced is smoother than 100fps not vsynced).

I dont know if its the placebo effect that makes you feel its smoother or whats going on, but i've heard so many CS noobs say that you need 100fps to be accurate in that game while using a normal 60hz 20ms tft panel ;D

Remember kids, use Vsync !

[imbaczek]

because if your game runs on 100fps, input is (should be!) processed at 100fps even though the display refreshes only 60 times per second.

[Tobi]

And if the game does it right, it limits rendering to the monitor refresh rate and spents the entire extra time to loop through the input routine.

Since the input routine is usually very cheap, this could result in over 9000 input FPS @ 60 rendering FPS, while otherwise it'd only be able to do 61-65 combined FPS

[Neddie]

What ever you say, there is no way you could in any way feel or hear FPS rating higher then what your display is capable of updating.

So a typical TFT screen only updates 60 times a second, so 60fps is exactly as smooth as 125FPS. (Well, probably 60fps vsynced is smoother than 100fps not vsynced).

I dont know if its the placebo effect that makes you feel its smoother or whats going on, but i've heard so many CS noobs say that you need 100fps to be accurate in that game while using a normal 60hz 20ms tft panel ;D

Remember kids, use Vsync !

All true.

[Zpock]

That's why hardcore FPS players typically try their hardest to get their grubby mittens on the last dying breeds of oldskool CRTs?

[s3cco]

What ever you say, there is no way you could in any way feel or hear FPS rating higher then what your display is capable of updating.

So a typical TFT screen only updates 60 times a second, so 60fps is exactly as smooth as 125FPS. (Well, probably 60fps vsynced is smoother than 100fps not vsynced).

I dont know if its the placebo effect that makes you feel its smoother or whats going on, but i've heard so many CS noobs say that you need 100fps to be accurate in that game while using a normal 60hz 20ms tft panel ;D

Remember kids, use Vsync !

hmm... no
first of all:

if your game runs on 100fps, input is (should be!) processed at 100fps even though the display refreshes only 60 times per second.

secondly, do yourself a favour and DON'T use vsync because, even though vsync makes your game as (visually) smooth as your monitor can afford, it causes big input lag.
The best thing you can do is experiment with different fps caps above the monitor refresh rate to reduce tearing while keeping it at least in the 100-120 range (for comparison, usb polling is 125 Hz and many people overclock it for better mouse response)

This said, if you have a "normal 60hz 20ms tft panel" you're screwed by definition, noobs and pros alike will pwn you as long as they have a decent monitor, and you will fuck up your eyes as well :)

This doesn't mean you can't enjoy playing a fps even with a crappy monitor, a 15 year old ball mouse and a 56k connection, but once you get used to smoothness you'll find it hard to readjust yourself to lag (which is one of the reasons some people still play Quakeworld)

[Ixoran]

@ Vadi

For about 120 years, the accepted figures for mean simple reaction times for college-age individuals have been about 190 ms (0.19 sec) for light stimuli and about 160 ms for sound stimuli (Galton, 1899; Fieandt et al., 1956; Welford, 1980; Brebner and Welford, 1980).

But you must also understand reaction time is directly linked with perception and prediction.
IE The better you can predict the forcoming situation the faster you can react to it. For example, If you know that in exactly one second a man will jump from a corner, and his head will be .5 meters from the wall of that corner, and 2 meters above the ground, you can shoot him in the face with some incredibly fast reaction time, because only your trigger finger has to really react, not your whole arm to align the shot, etc.

Very interesting article.

Hmm, You guys are making me want to keep my CRT as a primary monitor and just get an LCD as a secondary.

[SwiftSpear]

What ever you say, there is no way you could in any way feel or hear FPS rating higher then what your display is capable of updating.

So a typical TFT screen only updates 60 times a second, so 60fps is exactly as smooth as 125FPS. (Well, probably 60fps vsynced is smoother than 100fps not vsynced).

I dont know if its the placebo effect that makes you feel its smoother or whats going on, but i've heard so many CS noobs say that you need 100fps to be accurate in that game while using a normal 60hz 20ms tft panel ;D

Remember kids, use Vsync !

In CS the graphical FPS match the calculation FPS. Like I say, while I can't see the difference between 60 and 100 FPS, the game feels different, because it's running fewer calculation paths per second. This is most notable with bhopping. Both scripted and mousewheel, a 100 fps bhop is alot more responsive than one at 60 FPS. Plus alot of FPS players till use CRT for the improved hertz age.

I never use Vsync. If you have Vsync on, and your PC is only capable of running at 59 frames per a second on the scene, it will bump you down to 30 FPS, which is VERY noticeable VS 60 FPS in most games.

[Vadi]

@ Vadi

For about 120 years, the accepted figures for mean simple reaction times for college-age individuals have been about 190 ms (0.19 sec) for light stimuli and about 160 ms for sound stimuli (Galton, 1899; Fieandt et al., 1956; Welford, 1980; Brebner and Welford, 1980).

But you must also understand reaction time is directly linked with perception and prediction.
IE The better you can predict the forcoming situation the faster you can react to it. For example, If you know that in exactly one second a man will jump from a corner, and his head will be .5 meters from the wall of that corner, and 2 meters above the ground, you can shoot him in the face with some incredibly fast reaction time, because only your trigger finger has to really react, not your whole arm to align the shot, etc.

Very interesting article.

Hmm, You guys are making me want to keep my CRT as a primary monitor and just get an LCD as a secondary.

Well, thanks, just goes to help show that 5ms when the reaction time is 190ms don't matter much like I was saying.

[SwiftSpear]

@ Vadi

For about 120 years, the accepted figures for mean simple reaction times for college-age individuals have been about 190 ms (0.19 sec) for light stimuli and about 160 ms for sound stimuli (Galton, 1899; Fieandt et al., 1956; Welford, 1980; Brebner and Welford, 1980).

But you must also understand reaction time is directly linked with perception and prediction.
IE The better you can predict the forcoming situation the faster you can react to it. For example, If you know that in exactly one second a man will jump from a corner, and his head will be .5 meters from the wall of that corner, and 2 meters above the ground, you can shoot him in the face with some incredibly fast reaction time, because only your trigger finger has to really react, not your whole arm to align the shot, etc.

Very interesting article.

Hmm, You guys are making me want to keep my CRT as a primary monitor and just get an LCD as a secondary.

Well, thanks, just goes to help show that 5ms when the reaction time is 190ms don't matter much like I was saying.

In CS, the best players master the game to obscene quantities. 190ms is a cold reaction time. IE, if my friend is holding a ruler up, and he drops it, how immediately can I catch it. I know my friend will drop the ruler, but I don't know when, or how long I will wait.

CS players do a plethora of things to improve their reaction times. First of all, the best players have just practiced their reaction time to the point where it's above the normal human average. Second, CS players constantly listen for the location of their opponents so they don't have to be cold on reaction when his head pops around the corner, third, CS players master cyclic reaction loops, so putting the crosshairs on your opponent's head is a matter of instinct, as opposed to conscious reaction. Now pair that with the fact that in CS, if my opponent gets a headshot on me before I get a headshot on him, that's it, he wins. Since we're likely to see eachother at precisely the same moment in CS, if he can shoot my head in 190ms, but I take 195ms to shoot his head. I still loose, even though the differences in our reaction times were 2%, fundamentally, one shot still executed before the other did. Pair this with FPS, where if my monitor is running capable of 100FPS, vs someone else's which is running capable of 60, worst case scenario for that person, I'm getting an extra 0.006 seconds of time to react. So not only is his LCD adding 5MS of response time (which also will effect coordination of mouse action to screen action negatively) it's potentially adding a total of 0.011 second display delay. that's 6% of a human's cold reaction time, which is significant, especially since, as I noted, cold reaction time is often significantly lower than warm reaction time.

[Hobo Joe]

CS players master cyclic reaction loops, so putting the crosshairs on your opponent's head is a matter of instinct, as opposed to conscious reaction.

This is definitely the biggest thing. While your actual 'reaction' time is 150ms or so, the time it takes for your instinctual reaction to put your crosshair on the other player and get a shot off can all happen in less than the time it takes for you to even realize what you're doing.

Oftentimes in CS/CSS I'll kill another player and not realize I've done it until after they're dead, because it's all instinctive and instantaneous reaction.

[SwiftSpear]

CS players master cyclic reaction loops, so putting the crosshairs on your opponent's head is a matter of instinct, as opposed to conscious reaction.

This is definitely the biggest thing. While your actual 'reaction' time is 150ms or so, the time it takes for your instinctual reaction to put your crosshair on the other player and get a shot off can all happen in less than the time it takes for you to even realize what you're doing.

Oftentimes in CS/CSS I'll kill another player and not realize I've done it until after they're dead, because it's all instinctive and instantaneous reaction.

Not instantaneous, per say, but in terms of sight of model to incident of shot speed, I'd say likely SIGNIFICANTLY under 190ms. Probably still significantly over 11ms, but still...

[reivanen]

In CS the graphical FPS match the calculation FPS. Like I say, while I can't see the difference between 60 and 100 FPS, the game feels different, because it's running fewer calculation paths per second. This is most notable with bhopping. Both scripted and mousewheel, a 100 fps bhop is alot more responsive than one at 60 FPS. Plus alot of FPS players till use CRT for the improved hertz age.

I never use Vsync. If you have Vsync on, and your PC is only capable of running at 59 frames per a second on the scene, it will bump you down to 30 FPS, which is VERY noticeable VS 60 FPS in most games.

30fps vs 60fps is clearly noticeable, i give you that. But you have to yet somehow prove that you could feel the difference of a 10ms physics update interval compared to 16,7ms.

[BoredJoe]

In CS the graphical FPS match the calculation FPS. Like I say, while I can't see the difference between 60 and 100 FPS, the game feels different, because it's running fewer calculation paths per second. This is most notable with bhopping. Both scripted and mousewheel, a 100 fps bhop is alot more responsive than one at 60 FPS. Plus alot of FPS players till use CRT for the improved hertz age.

I never use Vsync. If you have Vsync on, and your PC is only capable of running at 59 frames per a second on the scene, it will bump you down to 30 FPS, which is VERY noticeable VS 60 FPS in most games.

30fps vs 60fps is clearly noticeable, i give you that. But you have to yet somehow prove that you could feel the difference of a 10ms physics update interval compared to 16,7ms.

In quake 3 there is a massive difference between 60fps and 125fps. For starters, when jumping you jump slightly higher for longer at 125fps because there are more updates at the peak of the jump. When making big jumps, you get to the platform you're jumping to very slightly faster, meaning you'll be slightly higher and have a better chance of making the jump.

When you're strafe jumping, the success of the jump relies on the precise movements you make with the mouse and keyboard between landing and jumping again - a period which decreases the faster you're moving. At low fps, you simply can't pull it off as well. When you're strafe jumping, you can easily end up moving 2-3 times as fast as you normally do in quake, so yes there is a massive difference.

[imbaczek]

vq3 is a bad example because framerate-dependent physics is _always_ a bad idea, and that's why CPMA has done it right.

of course, 125Hz input resolution comments still apply, especially for strafejumps.

[DemO]

What ever you say, there is no way you could in any way feel or hear FPS rating higher then what your display is capable of updating.

So a typical TFT screen only updates 60 times a second, so 60fps is exactly as smooth as 125FPS. (Well, probably 60fps vsynced is smoother than 100fps not vsynced).

I dont know if its the placebo effect that makes you feel its smoother or whats going on, but i've heard so many CS noobs say that you need 100fps to be accurate in that game while using a normal 60hz 20ms tft panel ;D

Remember kids, use Vsync !

All true.

Uh sorry but no. Vsync causes noticable input lag in fps games, which pretty much anyone who plays fast paced fps competitively with mouse and keyboard will tell you.

I can notice the difference in frame rates up to about 60 to 70fps, after which theres no noticable visual difference. Usually its only significant when making extremely fast mouse movements (flick shots for example) or in any game that's inherently fast paced.

And on the topic of response times, there are plenty to take into consider outwith your monitor when gaming. The standard refresh rate for the tracking on an optical/laser mouse is 125Hz for example (8ms), but I and many others will say that upping it to 500Hz (2ms) DOES make a minor, but nevertheless significant impact on "input lag" - or at least the brains perception of it - when playing fast paced FPS games.

In RTS games the difference between 40 and 60 fps solid is quite noticable in terms of the mouse cursor. You can see it much more fluidly and for some reason it feels more accurate at 60fps.

[Crayfish]

Those figures for reaction times are probably averages from a number of study participants. Not sure the ruler test is the best way to assay time to click either.

Try this one to find your time http://cognitivelabs.com/mydna_speedtestno.htm

Mine's not great - 248 ms

I know it's possible to improve with training though. Some martial arts include reaction time training drills which are quite effective.

[SwiftSpear]

I averaged 200 after a couple try of that test. That test is definitely testing cold reaction time though... and a keypress of that nature is going to add a small amount to your reaction speed as well. My best attempt got as low as 173.

Like I say, there are a plethora of sound and visual cues in FPS games, often time a reaction is not cold, it's warm. If I know where my opponent is going to be, and I know when I'm about to see him, I don't have to wait on my reaction time to kick in. Ask a drummer to tap along to an existing beat, you will find virtually no delay in his reaction from the time a beat starts to the time his tap hits, he has anticipated the beat before it comes because he has context to work with. The reality is, we are talking about a situation in which you aren't necessarily restricted by your reaction time. In those cases that 5ms delays can be significant. Probably not game determining, but they will have an effect.

[Vadi]

Keep in mind your ping too (which is measured in ms. And we're talking about 5ms here).

[Hobo Joe]

Keep in mind your ping too (which is measured in ms. And we're talking about 5ms here).

It is compensated.

http://developer.valvesoftware.com/wiki ... mpensation
http://developer.valvesoftware.com/wiki ... mpensation