flood's input lag measurements

[Dragonheart]

Sorry, my inattention

BTW, the latest 344.60 version gives the same responsive mouse feeling as 340.52. Confirmed by another player.

[flood]

np this thread is a complete mess
i'll test 340.43 vs 340.52 at lower fps tonight..

where do people discuss this driver/inputlag thing?

[Dragonheart]

where do people discuss this driver/inputlag thing?

Well, I came from some Russlan speaking overclockers forum, so I doubt it would be usefull for you
Anyway, any chances on results of comparing 340.52 with other versions on lower FPS? New 344.75 feels even worse than 344.60, while giving exactly the same FPS for me in a timedemo. So it would be interesting to see some testings with it.

Also, which settings in CS GO do you use? I set all maxed including AA (more static image makes it easier for aim for me) x16q CSAA, while I noticed that disabling AA completely gives a better response feelings. It may only be connected with FPS increase from ~340 to ~550 in my case, but since the actual frame times with such fps differ by only 1-2ms, I shouldn't feel the difference as much as I can feel. Also, I'm using Raw Input for the mouse which makes it independent from FPS, so it may affect the input lag measurements too somehow.

BTW, I've checked different prerender limit settings in a timedemo. First run was ignored, after every settings change the game was restarted. Here're the results:

0: 232.80 232.63 232.70
1: 388.62 388.39 388.50
2: 349.71 352.70 343.30
3: 344.30 347.20 349.90
4: 345.10 352.40 349.10
Auto: 346.11 344.69 348.70

344.75 driver has the same results. For emulating 0 setting I've used D3D9 Antilag library.

What's REALLY strange to me is that 1 frame limit gives the best FPS while it should give the lowest (not counting the emulated 0 buffer). 2-4 are almost the same (probably because in my case VGA is the bottleneck and there's no much time for the CPU to prerender anything). The 40 fps difference is really a big difference. So I suppose something doesn't work right. And I'm curious if there's a difference in input lag between 1 and other values What if 1 doesn't work like it's supposed too giving an increased buffer limit and a higher input lag?

[Dragonheart]

same stupid jokes year by year

[flood]

where do people discuss this driver/inputlag thing?

Well, I came from some Russlan speaking overclockers forum, so I doubt it would be usefull for you
Anyway, any chances on results of comparing 340.52 with other versions on lower FPS? New 344.75 feels even worse than 344.60, while giving exactly the same FPS for me in a timedemo. So it would be interesting to see some testings with it.

wups forgot about that.
i have an hour or so and ill try it now

Also, which settings in CS GO do you use? I set all maxed including AA (more static image makes it easier for aim for me) x16q CSAA, while I noticed that disabling AA completely gives a better response feelings. It may only be connected with FPS increase from ~340 to ~550 in my case, but since the actual frame times with such fps differ by only 1-2ms, I shouldn't feel the difference as much as I can feel. Also, I'm using Raw Input for the mouse which makes it independent from FPS, so it may affect the input lag measurements too somehow.

i use 4x msaa and it doesn't affect the measurements. ive never tried csaa

[flood]

btw what's the history of the
A B C
B
C
thing?


nvidia-branded gtx 970 at stock settings, 344.75 driver.
my standard csgo settings (1024x768@150hz, multicore off, raw input, all graphics settings at minimum except 4x msaa and trilinear filtering)

measurements in local server, custom map
framerate: 1700 +- 100fps

4,3,4,4,4ms


measurements in my dedicated server, dust2 with 9 bots running around, "hidehud 4" to disable hud
framerate: 400 +- 20fps
6,6,6,6,5,6,5,6,5,5ms

i have no idea how such low variation is possible. more on this later.

dedicated server, de_marquis, graphics settings maximum (high/very high on everything, 8x msaa, 16x af, multicore off, fxaa off), full hud
framerate: 240 +-20fps
7,10,12,7,9ms

switched to my default graphics settings, also with hud on
framerate: 280+-20 fps
10,7ms


in 344.75 there isn't prerendered frames, only "virtual reality prerendered frames"
i dont think it affects anything but lets try to make sure

again, de_marquis with max graphics:
9,8,10ms

so that one doesn't affect anything.

i can't test earlier driver versions than 344.x with gtx 970 so ill have to swap back to my gtx 460 to try the 340.x ones

but i gotta go now... ill try gtx460+340.x drivers tomorrow morning, assuming that i wake up before noon ;p

[Dragonheart]

in 344.75 there isn't prerendered frames, only "virtual reality prerendered frames"
i dont think it affects anything but lets try to make sure

I definitely see a Maximum pre-rendered frames option in 344,75 just like before aside from virtual pre-render frames option.

sorry lol, its for fun,. i hope to see where i can find the discussion, i can even translate the page and understand at least something

Well, the whole driver discussion topic is here but it's hard to find some input lag discussions now and there was no real measurements of lag anyway so this thread is pretty useless

[flood]

what card do you have?
it's entirely possible that some issues are specific to certain card architecture +driver combos

[Dragonheart]

what card do you have?
it's entirely possible that some issues are specific to certain card architecture +driver combos

GTX 770 + win 7 x64 sp1 with latest updates
The other guys that also reported the same lag problem with drivers newer than 340.52 have the same architecture - gtx660 and gtx770 too, so it may be the point..

[flood]

blah i'll test gtx 460 with various drivers when i get back on monday.

honestly it's a little tedious to do and so far there have been no unexpected results