Timer Resolution - 0.497 ms

Everything about latency. Tips, testing methods, mouse lag, display lag, game engine lag, network lag, whole input lag chain, VSYNC OFF vs VSYNC ON, and more! Input Lag Articles on Blur Busters.
Slender
Posts: 573
Joined: 25 Jan 2020, 17:55

Re: Timer Resolution - 0.497 ms

Post by Slender » 06 Feb 2023, 00:22

useplatformtick no(deletevalue) always show 0.499 0.496 0.497 or like that.
use 0.5ms timer not recomend on new windows build with useplatformtick yes, it decrease perfomance and highest dpc.
sometimes usetick no show 0.5ms on z690/z790 d4 because this platform use undisable spread spectrum in cpu.

vhagar
Posts: 10
Joined: 27 Sep 2022, 05:10

Re: Timer Resolution - 0.497 ms

Post by vhagar » 07 Feb 2023, 09:00

Slender wrote:
06 Feb 2023, 00:22
useplatformtick no(deletevalue) always show 0.499 0.496 0.497 or like that.
use 0.5ms timer not recomend on new windows build with useplatformtick yes, it decrease perfomance and highest dpc.
sometimes usetick no show 0.5ms on z690/z790 d4 because this platform use undisable spread spectrum in cpu.
Thank you for sharing your input. Yes, I am running on a ROG STRIX Z690-A GAMING WIFI D4 - ASUS, and it still randomly shows 0.5ms with useplatformtick no—i generally reboot until it shows 0.497ms with timer resolution set to max.

Slender
Posts: 573
Joined: 25 Jan 2020, 17:55

Re: Timer Resolution - 0.497 ms

Post by Slender » 07 Feb 2023, 09:11

vhagar wrote:
07 Feb 2023, 09:00
Slender wrote:
06 Feb 2023, 00:22
useplatformtick no(deletevalue) always show 0.499 0.496 0.497 or like that.
use 0.5ms timer not recomend on new windows build with useplatformtick yes, it decrease perfomance and highest dpc.
sometimes usetick no show 0.5ms on z690/z790 d4 because this platform use undisable spread spectrum in cpu.
Thank you for sharing your input. Yes, I am running on a ROG STRIX Z690-A GAMING WIFI D4 - ASUS, and it still randomly shows 0.5ms with useplatformtick no—i generally reboot until it shows 0.497ms with timer resolution set to max.
https://www.reddit.com/r/ASUS/comments/ ... mer_issue/

masneb
Posts: 239
Joined: 15 Apr 2019, 03:04

Re: Timer Resolution - 0.497 ms

Post by masneb » 07 Feb 2023, 18:54

Someone needs to test this stuff with reflex analyzer. There are posts stating that .5ms gives the lowest input latency. I guess that's part of why there is absolutely no clear answer on HPET and the timers, because it seems to change based on hardware and versions of windows.

I play a lot of OW and had synthetic timers off due to this post, just turned them back to default and tscsyn because of the above post to give it a try.

https://us.forums.blizzard.com/en/overw ... -11/652137

It all feels different and I really don't know if it's better or worse, all I can say is it's different. Even if input latency is lower, that doesn't mean you're more accurate as well as input latency isn't the consistency of the polling data nor the accuracy of the data in the stream (if error is produced by your mouse), nor how the engine processes said data.

And that's putting aside that it might and probably is effecting how the engine processes data in the first place, completely putting aside peripheral input. So a input analyzer might not even show anything. Windows is a giant steaming pile and there aren't any tools to diagnose this stuff besides input latency analyzer and FPS measurement tools (RTSS).

Slender
Posts: 573
Joined: 25 Jan 2020, 17:55

Re: Timer Resolution - 0.497 ms

Post by Slender » 12 Feb 2023, 17:21

masneb wrote:
07 Feb 2023, 18:54
Someone needs to test this stuff with reflex analyzer. There are posts stating that .5ms gives the lowest input latency. I guess that's part of why there is absolutely no clear answer on HPET and the timers, because it seems to change based on hardware and versions of windows.

I play a lot of OW and had synthetic timers off due to this post, just turned them back to default and tscsyn because of the above post to give it a try.

https://us.forums.blizzard.com/en/overw ... -11/652137

It all feels different and I really don't know if it's better or worse, all I can say is it's different. Even if input latency is lower, that doesn't mean you're more accurate as well as input latency isn't the consistency of the polling data nor the accuracy of the data in the stream (if error is produced by your mouse), nor how the engine processes said data.

And that's putting aside that it might and probably is effecting how the engine processes data in the first place, completely putting aside peripheral input. So a input analyzer might not even show anything. Windows is a giant steaming pile and there aren't any tools to diagnose this stuff besides input latency analyzer and FPS measurement tools (RTSS).
try to use:
bcdedit /set useplatformtick yes +1ms default timer
also, set Legacy RTC timer in your motherboard via grub for stabilize dpc latency.

masneb
Posts: 239
Joined: 15 Apr 2019, 03:04

Re: Timer Resolution - 0.497 ms

Post by masneb » 18 Feb 2023, 00:02

I've tried all the snakeoil fixes, was mentioning that there is no qualitative testing of things like this and there needs to be and that upon testing them subjectively it's hard to gauge if they're better or worse. I think most people are in this boat.

Post Reply