Blur Busters Forums

running w10 1909, and a 2080ti

I have a setup with a PG279Q running at 120hz and a XL2746S at 240hz both using 8K cert displayport on same graphic card (2080ti) ,

I noticed that when I test my reaction time (human benchmark site) on the benq and both monitor are enabled (not cloned just 2 desktop) my average is about 168ms.

As soon as I disable the 120hz monitor via nvidia control panel, I average 158ms or about 10ms less
It really doesnt seem to be a placebo as I reproducted this multiple time.

anyone noticed something similar ?

Yes.

Different-Hz multimonitor has this problem.

This is widely known at Blur Busters that multimonitor adds latency interference --

Microsoft Windows does not support truly multithreaded independent-Hz dwm.exe compositors -- it synchronizes to whatever monitor is getting the most activity, or the primary monitor.

To solve this, always run games in full screen exclusive mode on the primary monitor. Not borderless windowed, not windowed -- since the other monitors will interfere (especially if a window is open on both of them).

If you want lowest concurrent multimonitor lag, all monitors must be running the same refresh rate with the same Custom Resolution Utility (CRU) numbers -- identical porch numbers, identical sync numbers, identical total numbers, identical pixel clock numbers. This allows all monitors to run synchronized, bypassing compositor problems (stutters/lag).

Just tested it and I don't notice such thing running 1080p 210hz and 1440p 144hz screens.

PixelDuck87 wrote: ↑

27 May 2020, 19:09

Just tested it and I don't notice such thing running 1080p 210hz and 1440p 144hz screens.

in order to notice it you need to disable the monitor not just power it off, in nvidia control panel you need to untick it on the setup mutiple display section.

for me there is a clear difference of about 10ms which fall between 1 and 2 refresh cycle of the 120hz monitor.

what version of windows are you running ?

PixelDuck87 wrote: ↑

27 May 2020, 19:09

Just tested it and I don't notice such thing running 1080p 210hz and 1440p 144hz screens.

If you run two www.testufo.com windows on both screens (even in 2 different web browsers, FireFox and Chrome) you'll notice one of them microstutter out of sync.

Fortunately, at 210Hz, your probable stutter/latency error is 1/210sec (4.8ms) injected into the 144Hz refresh cycles with a halftime of 0.5/210sec and a latency jitter of +/-2.4ms. This may be below direct human detection thresholds during unsynchronized VSYNC OFF situations (since VSYNC OFF microstutter will somewhat hide the discordant-Hz multimonitor stutter, especially if your mouse is at only 400dpi). But may be applicable to a Millisecond Matters situation, where the millisecond is more important.

Fortunately, this problem disappears if you use Full Screen Exclusive on the primary, especially if not a single pixel is animating on the 2nd monitor (secondary). Then it behaves like single-monitor latency.