So just a update on things here, I've tried a ton of different things and this seems to be a very complicated issue. It seems this specific test is good for detecting micro stuttering and problems with frame delivery, not just for testing monitors, and as such I was able to test a plethora of different things out to figure out which works best. Using a camera to slow down what you see and being able to see a rather static output is along the lines of something FCAT can do.
I'm using w10 x64, 2080, and a 8700k. Testing was done in Chrome, similar results happened in Firefox. I stream so I use a cloned desktop display with a secondary display as an extended monitor.
My original setup I took those screenshots with was a
1x extended desktop on a 1680x1050 60hz monitor off of the 8700k integrated graphics
1x mirrored 1920x1080 240hz PG258Q and a Magewell capture card off of the 2080. Both are operating at 240hz.
First I'll state after removing the extra monitors the PG258Q had no problems with the test. The weird font problem definitely seems to be a residual issue from when I first sent my monitor in, so that's not fixed.
To weed out the integrated, first thing I did was try out a different monitor and I removed my Magewell. Both a VG248QE and my 1680x1050 have the same issues. Going between extended off of integrated and extended on my 2080 resulted in less yellow error messages. Since there isn't a metric built into the test for recording stutters over time I had to subjectively gauge how often the errors appeared on the screen. An extended desktop didn't cause red errors to pop up at all.
Having everything on the same graphics adapter results in less stuttering then splitting between integrated and dedicated.
Further looking at this, it seems having ANYTHING graphically updating on my extended monitor, whether it's on integrated or on the 2080 results in stuttering. I normally have pingplotter open and this definitely triggers stutter warnings in the test. I have a few apps open on my secondary screen, that being one of them. Steam friends also does this to the point that the test can't sync unless it's in fullscreen (green arrows) and then it will occasionally pop yellow messages.
Refreshrate matching between all connected devices (in the case of the test above VG248QE 120hz, PG258Q 120hz) completely resolved the syncing problem, so I definitely could see this being a issue with Windows and differing refresh rates on connected monitors. I couldn't operate my monitors at 120hz off of integrated, so whether or not refreshrate matching works for that remains to be seen. Windows used to have a bug where in if you had a video playing on an extended monitor it would cause the main one to stutter if it was a high refreshrate display. It seems they only partially fixed the issue with their WDDM update in the 2017 fall creators update.
If the PG258Q is 240hz and the VG248QE is 120hz the issues persist. The desync seems to get worse the bigger the differentiation between hz as well, running the VG248QE at 60hz for instance.
It does not seem to matter if all devices are displayport or mix of displayport and hdmi.
If I have the same applications open and the same layout, including an extended display with the apps on the main monitor behind the browser all open and running the stutters don't happen. So it's not CPU related.
Now is where the real fun begins - mirroring. Cloning monitors also causes all sorts of issues. I did a few different tests here. The tool will not attempt to sync higher then the lowest refreshrate cloned display.
PG258Q and VG248QE-
Cloning both displays at refreshrate matched 120hz caused no issues, the tests worked perfectly fine. Occasional yellow stutters would show up.
Cloning at 120hz VG248QE and 240hz PG258Q worked, the test only operated at 120hz. I choose 120hz for the VG248QE in these tests just to weed out the idea of indivisible numbers causing issues. Going back to my original pictures in this thread artifacts started appearing regardless of the test syncing properly. Not half drawn frames, but rather color mismatch issues, which can be seen in the original photos and here:
PG258Q and Magewell HDMI-
Cloning displays at 240hz resulted in a occasional yellow stutters, but it synced fine. However when going into fullscreen the test resulted in major stuttering or it would not vsync. It would sync up for maybe 1-2s or so before going red.
The PG258Q and the Magewell at 120hz sync'd fine in normal mode, but going into fullscreen after a few seconds of working fine resulted in the unable to vsync or major stutters detected once again. Considering they're both at 120hz and this worked fine with the VG248QE and the PG258Q one could surmise there is a issue with cloning a DP and a HDMI monitor.
PG258Q at 240hz and Magewell at 120hz resulted in the same as the VG248QE at 120hz and the PG258Q at 240hz. Weird color inconsistency, but no torn frames as was seen in the above pictures. Unlike the above two refreshrate locked tests, this one didn't fail on fulscreen.
PG258Q 240hz, Magwell 240hz, and VG248QE 120hz-
I was able to get the test working properly only when the VG248QE and Magewell were in extended mode with nothing on the desktop on either of them that moved. Going back to the earlier testing with integrated and extended anything that moves on an extended desktop causes stuttering. The more it's animated or moves around, the more stutters it causes, from yellow all the way to red. With my extended desktop fully loaded I was unable to get the test to work in normal mode. The stuttering errors get worse if I hooked up the VG248QE to my integrated or if I ran it at 60hz.
However, in extended with the desktop loaded up if I went fullscreen on the PG258Q with the test it would sync up and work fine.
Now if I try to clone the Magewell and the PG258Q and leave the VG248QE extended the test gives a vsync error in normal mode and in fullscreen mode. The framerate never hits 240. I'm not sure what I did to take the original pictures, but it's no longer working.
Sooo, why such a long winded post? I'm very sensitive to stuttering in games and it seems like this test is good for figuring out weird Windows configurations that Windows doesn't like. Do they all work and I have a screen on? Yes, however they don't all work the same.
For instance I wont be using integrated graphics anymore. I originally used it as I thought it would help offload some of the workload from my main GPU, it ended up just making things worse.
Refresh rate locking seems to help immensely, as long as all devices are the same refreshrate. The bigger the difference between the monitors, the worse the stuttering gets and the test wont even work properly. 240hz cloned didn't even work properly in the test in fullscreen for whatever reason, given how new 240hz is, Windows or Nvidia may still very well have some bugs. This could also be because the PG258Q is DP and the Magewell is HDMI.
Loading up your extended desktop with things very much impacts the experience on your primary monitor regardless of how little load it's putting on your CPU/GPU. There is some sort of weird WDDM bug that still exists regarding animated content. I'm sure no one puts anything on a second monitor that does anything though. Just static text.
This is something that mainly pertains to people who stream (such as those with capture PCs), but also to anyone who uses a secondary monitor which doesn't operate at the same refreshrate as their primary monitor (which is pretty much everyone).
Even though I can run my PG258Q in Gsync mode while cloned with my Magewell at 240hz, it does make me wonder if it's even doing anything or working properly.
This is all based on the premise that this test does what it's supposed to and isn't really buggy when it comes to cloned monitors, that in and of itself might just make this whole thing pointless.
Enjoy.
