Possibly Reduce Input Lag, System Latency, and Stutters, and Possibly Raise FPS

[Spinelli]

What is Timer Resolution?
How did this application get created?
Answered here --> http://www.lucashale.com/timer-resolution/

The free v1.2 works perfectly with Win 7, 8, and 8.1. Also, it sets the timer resolution to 0.500ms (which is what it should be for best results) rather than only 1.000ms like the chart says.

Just download the free v1.2, open it, click "maximum", see the number change from either 15.625 or 1.001, to 0.500, and minimize the program. Make sure to not actually close it though.


NOTE: Some games can request a lower timer from Windows, for example 10.000ms or even 1.000ms - and I'm betting most sims, especially ISI-engine based ones, do this - but this program will ensure you're running at just 0.500ms.


- Some people get framerate increases, some get quite large framerate increases in certain areas of CryEngine games like Crysis 3.

- It may reduce overall PC-system latency (great for people trying to run a server, or just overall PC and gaming use in general)

- It may reduce input lag (obviously great for simracing)

- Some people running servers have reported smoother online experiences due to the reduced latency

I have experienced slightly less input lag in some games with this program in the backround and have now made a habit of always running it - especially during gaming - for the past few years. In an absolute worst-case scenario it won't do any harm.


Some people's/gamers' experiences:

A.

According to http://www.shoot-em.com/forum/viewtopic.php?t=3981 my little program TimerResolution can be used to run a lag free counter strike server.

The internal windows timer is optimized for desktop applications and not for game servers.
This timer controls the time a thread gains to do it's work. By the default this interval is too long and you won't get more than 100 fps without lowering it.

(referring to a Counter Strike gaming server)

B.

C.

I’m told one of the uses for TimerResolution is to help reduce DPC Latency on Windows. This is supposed to improve performance of many games and applications.

I suspect that all that is really happening is that some poorly written drivers are getting a boost.

A great utility to check the DPC Latency on your system is http://www.thesycon.de/deu/latency_check.shtml, try it with and without TimerResolution running and see if it makes any difference on your system.

For a long technical write-up about all this stuff (way over my head) --> https://randomascii.wordpress.com/2013/07/08/windows-timer-resolution-megawatts-wasted/



P.S. Tim Wheatley from ISI immediately locked my thread about this on the ISI forums (yes, a technical PC program which may help simracers apparently cannot even be discussed there, not even just for the sake of testing and curiosity - unbelievable), and then banned me for an entire month just because I re-posted about it, so I am posting it here (RD is my 2nd most visited forum). I would love to hear people's feedback regarding the Windows timer resolution changes(which, in a worst case scenario, shouldn't do anything bad). I tried doing DPC latency tests using the program "LatencyMon" but I have no idea how to read all the values in that program. The program "Latency Check" (link in point "C") seems simpler to use/read but I haven't tried it yet.

 

[Spinelli]

Another positive result:

"Maximize your Windows timer resolution
Just keeping this little application open in the background will reduce DPC latency, improve responsiveness. Good workaround particularly for those who are stuck with an enforced HPET On.

I tested this on my workstation dual core who is using a lot of CPU time constantly. HPET is off.

DPC Latency is ~650us at 0.977ms timer resolution (default with my apps running).
DPC Latency is ~190us at 0.5ms timer resolution (maximized).

I now keep this app running constantly in the background, everything feel snappier.

On my gaming rig, HPET On (for this test purpose only):

DPC Latency is ~100us at 1ms timer resolution (default with DPC Latency Checker running, Windows default is normally 10ms)
DPC Latency is ~10us with some peak at 100us at 0.5ms timer resolution (maximized)

Entering a game normally bring timer to 1ms but you can improve it even more with this application running in background to 0.5ms. Depending the game and your system you may feel better responsiveness and less stuttering."

Those are some VERY, VERY positive results.

P.S. Ignore the whole "HPET" thing; don't worry about that.

 

[Eric Bergeron]

P.S. Tim Wheatley from ISI immediately locked my thread about this on the ISI forums (yes, a technical PC program which may help simracers apparently cannot even be discussed there, not even just for the sake of testing and curiosity - unbelievable), and then banned me for an entire month just because I re-posted about it, so I am posting it here (RD is my 2nd most visited forum).

not sure he like this software !
http://isiforums.net/f/showthread.php/24066-Lower-Input-Lag-Reduce-System-Latency

 

[Spinelli]

I know, really weird how hostile he became just for a program which may help some people. Really, really weird how it couldn't even be discussed there. There are a lot of very technical people on that forum - and I'm sure here at RD as well- and I would have loved to hear what others' results were, but he just freaked-out as if this whole Windows timer resolution thing and DPC latency improvements is all just "snake oil" or a virus or something. Very, very odd behaviour.

Anyways, enough about that. Has anyone done any tests? It clearly improves things while on the desktop with no other programs running since Windows runs at a snail-like 15.625ms, but I would love to see some technnical data from the following two scenarios:

1. RFactor 2 (or any other simracing title) player
2. RFactor 2 (or any other simracing title) user running a server


Make sure to not have any program running in the backround. For eg. I believe running Windows Media Player will lower the Windows timer resolution to 3.xxxms (a definite improvement over Windows' default 15.625ms).

 

[Eric Bergeron]

i tried but didn't noticed any difference !

 

[Spinelli]

If RFactor 2 itself requests a timer of 1.0 then I don't think lowering it down to 0.5 will make a huge difference. It may be extremely difficult to notice but an improvement may still be there.

Has anyone tried using any of the DPC Latency checker programs? What about admins running heavily loaded servers?


P.S. This is not some shady snake-oil or anything like that, it's just a regular part of the Windows operating system. I really hope people try to actually understand what this program does and how the Windows timer resolution works, especially with regards to DPC latency. AT THE WORST, setting Windows' timer resolution to 0.5 will simply do nothing rather than harm

 

[vee10]

Hi Spinelli, it's a very interesting subject. Do you have a DPC checker you recommend I could try.

 

[Spinelli]

There should be a link to two different DPC latency programs in my original post. One is for "LatencyMon" which I tried using but it had way too many values and I therefore had no idea what I was looking at. The other us "Latency Check" which seems simpler to use but I haven't tried it yet.

I saw a few guys on the ISI RF2 forums discussing LatencyMon and their test results with it (not regarding Windows timer resolution) so there are a few people there who definitely seem to understand all this stuff. Perhaps someone could simply post a link there to this thread for anyone interested in this topic (I imagine DPC Latency improvements would be a topic any gamer or server-admin would be interested, let alone ones of a "hardcore" racing sim).

 

[Robin]

I find many games already put the latency down to 1ms.
Having that extra half isn't going to do magic.

I've run this app for a few years but stopped after getting a new pc.
Don't notice I am missing something at this point.

 

[Spinelli]

I agree, but one thing that strikes me is the test results from my 2nd post. Having the Windows resolution timer @ 0.977ms resulted in 650us DPC latency while the DPC latency plummeted to just 190us when the program was used to set a Windows resolution timer of 0.500ms ("maximized") - almost a 3.5x improvement.

I'm lead to believe that having Windows' resolution timer set to 0.500ms ("maximized") must bring at-least small improvements even if they're too small too stand-out and make themselves noticed. Sort of like going from an 8ms input lag monitor to a 5ms one (just an eg.); the improvement will always be there in-effect even if one's senses can't notice it.

 

[Ari Antero]

I find also Tim`s behavior strange, however I dont know if you guys have noticed that ISI has edited lately rFactor2 http://rfactor.net/web/rf2/rf2help/ "What are the minimum specifications?" with this sentence:
"The better your machine is, generally the better our software will run. There are a few hardware solutions you should try to avoid, such as SLI, which increases latency. A single, powerful card, will usually run better in all games and simulations."

I do not think that it is fair that ISI edit specifications that suit to them after the release of the game and have reported this to NVIDIA and here is NVIDIA`s reply:
"Thanks for reporting this issue. I see some similar reports, and have passed along your information for further investigation. However, it's possible that the game itself is poorly optimized for multiple GPUs, in which case there wouldn't be much we could do on our end."

I thought that supported games from Nvidia like rFactor2 gives gaining a significant performance boost when running an NVIDIA SLI gaming PC and that all games work with NVIDIA SLI gaming PC, I thought also that latency isn't a big deal at higher framerates.

 

[Spinelli]

Ya Ari, I've read an official NVidia documentation for developers. It seems to point to SLI performance not being solely based in GPU drivers but game programming aswell.

Back to the Windows resolution timer, even though many programs set it to 1, the test results in my 2nd post show a massive DPC latency improvement going from 0.9xxms to 0.500 ("maximized"). Absolute worst case scenario, it simply doesn't do any harm but most likely brings slight improvements by lowering DPC latency even compared to what other games (eg. RF2). I therefore see no reason to not run the program, and no reason why it cannot be discussed in the ISI forums.

 

[Gijs van Elderen]

@Spinelli ,

Just a question:
Do you notice a CPU increase with the adjusted timer resolution?

I think it should if it's working. Because some processes are proccesed twice as much if you go from 1ms to 0,5ms.

 

[FredSmith]

Microsoft states the minimum is 1ms so anything below that is a complete waste of time. There's a reason Tim bans idiots.
https://msdn.microsoft.com/en-us/library/windows/hardware/dn265247(v=vs.85).aspx

 

[Gijs van Elderen]

Microsoft states the minimum is 1ms so anything below that is a complete waste of time. There's a reason Tim bans idiots.
https://msdn.microsoft.com/en-us/library/windows/hardware/dn265247(v=vs.85).aspx

At work we have PC's running programs with a timer resolution of 50μs. Click here
To have low response times, low latency at in and outputs. Sounds familiar?
With special I/O devices "ns" are possible. click here

The program runs on Windows 7.

I work on the hard-ware end. So don't ask me about details.

 

[Juergen]

Microsoft states the minimum is 1ms so anything below that is a complete waste of time. There's a reason Tim bans idiots.
https://msdn.microsoft.com/en-us/library/windows/hardware/dn265247(v=vs.85).aspx

Giving people names is much better?

 

[Spinelli]

Microsoft states the minimum is 1ms so anything below that is a complete waste of time. There's a reason Tim bans idiots.
https://msdn.microsoft.com/en-us/library/windows/hardware/dn265247(v=vs.85).aspx

- Is Tim your God? IIs he the most intelligent all-knowing person?

- Have I not shown you proof in test results showing an almost 3.5x improvement in DPC latency when going from a programs set 0.9xxms to using the tool to set it to 0.500ms ("maximized")?
HINT: Try reading post #2!!

- Have you not read the large article I linked to explaining all of this?

- Is that how you treat people who are trying to discuss & test possible game-improving software?

Is that ("idiot") what you call someone who is taking time and effort to spread information to others in order to try and help with a program that has shown proof of improvement from tests, and in a worse-case scenario will still not cause any harm?

If you're here to call someone an idiot for trying to help others, and do tests of completely legitimate, proven, non-harming software, then please stay out and go think about your behaviour because the type of actions you displayed are a prime example of what's wrong about society today.

On top of all that, and on a more personal note, I'd say that 50%-75% of my posts are regarding helping others, mostly regarding PC buying advice, hardware problems, software problems, and PC/GFX settings. I have also taken quite a bit of time to create threads about detailed graphics settings and explanations which have helped many, many people. So, please, don't attack the "good guy", go spread your evil & posinous hatred/energy somewhere else, or better-yet try changing - the world sadly already has enough people like you.

 

[Marc Collins]

I tried it. Made zero difference to the small micro-stutter issue I have since switching to an AMD card.

 

[Spinelli]

Marc, with regards to your issue, how are the stutters when your game's framerate, your screen's refresh rate, and your RF2 player.json file are all nicely matched-up?

 

[MarcG]

Spinelli please just give it up, yes you write some good threads with genuine advice but then you write these threads which lead only to confusion for the sake of absolutley no gain.
Not every thread you write is a God Send Fix for everyones machine, more often than not it's just another way for users to mess up their system and create problems they didn't have in the first place. What happens then? they go running to ISI saying it's their game at fault when in actual fact it's the Users tweaking.
Whilst these threads may help the minority they do not help everyone, as stated already to you it doesn't do Jack in Windows 8/8.1 and many people in Windows 7 see bugger all difference, if there was such a dramtric difference don't you think this "trick" would be more well known by now? (that rhetorical, don't answer that)

Finally I'm still amazed that you don't see why you got banned in the first place, it's blatantly obvious, ISI find these threads annoying...when your ban ends I'd advise you to not continute to post them.