flood's input lag measurements

[lexlazootin]

wow... i'm so stupid. pretend i didn't say anything.

Edit: Well you could do 2 tests drawing squares with the mouse jumping 6 pixels to the right, 6 down, 6 left then 6 back up. then doing it again but 3 pixels to the right twice, 3 pixels down twice... and checking if they land on all the same corners of the last test.

that would test accel in every direction.

[spacediver]

I don't think flood's test was measuring accel. It was meant to measure dropped samples.

[flood]

http://www.overclock.net/t/1545382/csgo ... ps-samples

for every useful post on ocn there are 5 that just talk about how things feel subjectively

[spacediver]

It'd be great to see a good double blind run.

[flood]

welp

[flood]

uhmm

anyway i just did 10 runs switching between rawinput 0 and 1 kind of randomly, underclocked cpu+gpu, bots kicked from server so that fps is stable at 90. i did these consecutively, not alt-tabing or doing anything in between. fps remained stable except during the first run where it dropped to 70 for few seconds (maybe affected 5% of the data in that run)

raw data: https://docs.google.com/spreadsheets/d/ ... 1426052600

averages in microseconds
run1, m_rawinput 1:17976.021
run2, m_rawinput 0 :16401.25
run3, m_rawinput 0:16337.279
run4, m_rawinput 1:17818.37
run5, m_rawinput 1:17953.225
run6, m_rawinput 0:16269.625
run7, m_rawinput 1:17812.031
run8, m_rawinput 0:16421.492
run9, m_rawinput 0:16373.09
run10, m_rawinput 1:17959.365

m_rawinput 0 is consistently faster by 1.5ms

this is pretty solid evidence that m_rawinput 0 and m_rawinput 1 do not behave identically in terms of absolute responsiveness.

[dmbr]

welp

Would you mind interpreting the graph for my puny brain?

I'm quite interested in the lag differences between m_rawinput and RInput.

[lexlazootin]

All the results in the graph are between 3000-6000 MICROseconds, which are 1,000,000th of a second. All the results being so close together shows that there pretty much no difference between any of the settings.

[dmbr]

All the results in the graph are between 3000-6000 MICROseconds, which are 1,000,000th of a second. All the results being so close together shows that there pretty much no difference between any of the settings.

Thank you

Just to be clear, this is measuring input lag, yes?

[lexlazootin]

yes