Blur Busters Forums

Does anyone has any experience with this little app ? Is there any benefit of using maximum timer resolution of 0.500 millisecond in comparison to 1.000 milliseonds which most games are using ?

Nope.

Why not ? Care to elaborate ? From the app description and by knowing what actually timer resolution in Windows does, then it would be logical that this can improve game/mouse responsiveness (maybe in a similar way as 1000Hz mouse does over 500Hz/125Hz mouse).

mello wrote:Why not ? Care to elaborate ? From the app description and by knowing what actually timer resolution in Windows does, then it would be logical that this can improve game/mouse responsiveness (maybe in a similar way as 1000Hz mouse does over 500Hz/125Hz mouse).

Because a difference of 0.5 ms is nothing. There will be no difference whatsoever, other then higher power consumption due to the timer interrupt firing twice as much.

But I'm pretty sure you figured this out by yourself, by now. Otherwise you wouldn't have asked

I've heard of people try to eliminate as many sources of latencies as possible, eliminating 0.5ms here, eliminating 0.3ms there, and another 1ms there. It eventually all adds up. By itself, it doesn't make a noticeable difference, but doing many things would.

Differential latencies can be a big thing -- you train for one latency, expecting a specific behavior, and the latency suddenly change -- it throws off aiming. A 5ms difference would be a 20 pixel overshoot at 4000 pixels/second 180 degree flick. (5/1000ths of 4000 = 20)

Also there are cascading effects possible. Improving a latency to a specific threshold, may suddenly improve other latencies by a bigger amount. More research would be needed.

interesting, so this app (the paid version at least) can allow your operating system to run at 2000 hz as opposed to 1000 hz?

Sounds like it might be worth experimenting with.

not worth using

I believe the only case where it would affect anything would be a game that forces a constant fps using a loop with Sleep

check the standard deviation of frametimes in a fraps benchmark to be sure.

flood wrote:I believe the only case where it would affect anything would be a game that forces a constant fps using a loop with Sleep

Sleep() uses whole milliseconds. So I don't think it would have much of an effect.

RealNC wrote:

flood wrote:I believe the only case where it would affect anything would be a game that forces a constant fps using a loop with Sleep

Sleep() uses whole milliseconds. So I don't think it would have much of an effect.

Busy wait loops on multimedia timers without sleeps, would be far more accurate. More power consuming, but for delays under 1ms, in certain cases it may actually lead to reduced stutters depending on how the frametime logic works, due to more consistent frame times at 0.5ms granularity. But most engines do not depend on this for frametimes during internal game-engine frame rate capping.

1ms random fluctuations in frame times means sudden jumps from 144fps (7ms) to 125fps (8ms) to 111fps (9ms) to 100fps (10ms). THIS is where better-than-0.5ms granularity is frequently needed, not for input lag, so sleeps are not usually used to wait for the next frame time. In theory, 0.5ms timer ticks could make accurate usleep() easier for the operating system to do power management at 0.5ms intervals (microsecond sleeping) since otherwise possibly energy-hogging busy wait loops are needed for accurate sub-1ms waits. On the other hand, the overhead of the higher frequency may outweigh this. All a matter more important for mobile devices.

TL;DR: it is not going to appreciably affect input lag in typical games, but one odd game may suddenly have fewer stutters if its frametime logic depends on this timer.

I have had this fix running for a while and I do not think it makes a noticeable difference but maybe it does for some engines. This video http://www.youtube.com/watch?v=_hcuYiqib9I claims it makes a big difference in crysis 3 if you are cpu limited(good chance in this game).
I don't know if it really helps but you could test out some games which are cpu limited and then see if it helps with frame rates. I will just leave it running since it is not a bother at all once working. Maybe some engines just implemented this wrong or overlooked it and they will perform better after this but I am not sure.