flood's input lag measurements

[flood]

i used the camera's player if you pause it, the wheel can be used for frame-stepping back and forth.

if you want to look on the computer though, well what you want is a video player that can frame step backwards.
see http://superuser.com/questions/45803/vi ... ck-feature
kmplayer should probably work, but i havent tried yet

what i do is place both mice on a white sheet of paper, illuminate the paper with a LED lamp (incandescent flickers at 120hz, and i dont have any nonflickering fluorescent lighting), and keep the active mouse near the edge so that there is only a sliver of white shown. the first frame where the sliver changes is the frame where the impact occurs.

i post some video frames later

[spacediver]

thanks.

why do you need two mice? Can't you just use your hand to nudge the mouse? Or a pencil or something?

[flood]

the dead mice isn't necessary but you need something is hard and somewhat durable, yet not so hard that it will damage my working mouse.

hand doesn't work because it's too soft. if i'm moving it at 1m/s, and my finger deforms by 3mm upon impact, that blurs the time of impact by 2ms.

and you want to make sure that the point of impact is low enough that the working mouse doesn't lift up

a pen could work i guess, but another mouse seems to be perfect for this

[Q83Ia7ta]

just picked up my casio ZR700, and am looking forward to following in your footsteps :p
any advice on what software to use to analyze the video files?

Try some VirtualDub mod or any player where you can see frame counter.
Also will be nice to see some better formatted text (not this mess by flood).

[LagBuster]

Hey, just wanted to mention how awesome this thread is.

Thanks a lot for your findings, Mr. flood. Really interesting stuff.

[flood]

Hey, just wanted to mention how awesome this thread is.

Thanks a lot for your findings, Mr. flood. Really interesting stuff.

thanks
if you haven't already, check out noacc's measurements on esreality: http://esreality.com/post/2640619/input ... -csgo-q3a/
he uses a different methodology but the results are still valid

Also will be nice to see some better formatted text (not this mess by flood).

yup i know... ill organize it once sometime when i have more data and conclusions but for now i'm just gonna use the first post to highlight the most important results

[Dragonheart]

flood, could you please test nvidia drivers - 340.52 which gives me the less amount of input lag comparing to some newer versions (344.11 was just awful)? At lest to prove that a simple user can/can't notice changes in input lag between drivers.
http://forums.blurbusters.com/viewtopic ... =40#p10937
I personally can't gain any close results to 340.52 while playing CS GO on other drivers. it becomes an infinite trying to compensate the input lag which leads to no aim and no signs of comfort while playing..

[flood]

will do

which graphics card do you have? i only have a gtx 460 and 970 to try

it probably is just in your head but it could be that specific graphics card/driver combinations have issues

[Dragonheart]

will do

which graphics card do you have? i only have a gtx 460 and 970 to try

it probably is just in your head but it could be that specific graphics card/driver combinations have issues

Nice!

I have gigabyte GTX770 2GB, Win 7 x64 SP1, HPET Enabled in bios and disabled in windows. Sad you are on win 8, it won't allow us to compare results in close conditions. CPU: 4770k @ 4.35GHz if it matters.
FPS in CS GO is always around 300 (limit is removed), so even a lag in 1 frame should give a very unnoticable input lag, but I feel what I feel. I tried playing for about a month with 340.43, but all I had - latest place and no aim. I've even lost 3 rank positions. After installing 340.52 - magic happened. Mouse feels ok again and I brought back my previous results like 1-2 place in competetive mode and guardian elite - distinguished guardian rank. I played for some time, then after installing 340.11 - problem with input came back. Same goes for the latest 344.48, but I feel my mouse slighty more responsive than with 340.11 (but still not even close to what I feel with 340.52).

AMD 7950 with prerender 0 via Radeon Pro gives me the same fluid and responsive experience on any driver version.

Some settings for nvidia driver I've changed:
Optimize for multiple displays: Single Display
Prerender limit: 1
Performance optimization: Maximum performance
Vsynk: off
Scaling: no scaling / Display
I have benq xl2411T with Instant mode enabled and AMA premium @ 144hz (enabling lightboost brings a noticable input lag too).

Settings in CS GO are also maxed out @ fullhd. I've only disabled fxaa / blur / vsynk. AA is 16xQ CSAA, so the fps is limited by VGA (which means that CPU has a room for buffering frames to prerender buffer, which is limited by 1 frame anyway.. just a thought).

PS. Some guys (battlefield players) on other overclockers forum confirmed that they have the best responsive mouse on 340.52.
PS2. Do you have creative x-fi audio like titanium or smth with a gaming mode?

[flood]

i have windows 7 too

note that i'm at home until sunday and all my stuff is at my dorm/apartment so i probably wont be measuring until then