TechnoRush wrote:I have a 29UM69G and I'm trying to extend the Freesync range of the monitor to take advantage of LFC, the stock range is 40 - 75.
I used Nvidia custom resolution to verify everything from 39 - 30 should display and it did just that, then used CRU 1.4.1 and edited the range and added a Freesync Data block for 30 - 75, restarted the system instead of the included .exe and enabled G-Sync in NVCP.
I then used Nvidia Pendulum to confirm my changes and... I'm getting flickering or what I call blackouts, specifically at 37 - 38 fps / hz and everything above and below that works as expected. G-Sync indicator is present and I can see G-Sync is working on the test pattern.
My monitor does not have Refresh rate monitoring on OSD.
Am I doing something wrong here or is it something else?
It might be a VRR range underclocking artifact.
You're underclocking your monitor severely at 37-38Hz, and sometimes underclocking causes blankout effects you see. When the LFC kicks in the problem solves itself, e.g. 28Hz means 28Hzx2 = 56Hz which is no longer underclocking. But once the frames stop doubling, e.g. 38Hz, you're trying to run native 38Hz, and that's the tiny underclock region you're hitting (38-39Hz).
You might find you need to do underclock/overclock both ends of the VRR range, e.g. 39Hz-78Hz -- to get the 2.0x you need for LFC. Normally you really need 2.4x for much more reliable LFC, as even 2.0x is still very tight.
Extending VRR range to 2x is necessary for LFC, but sometimes the act of extending this requires underclocking the low-end of VRR range or overclocking the high-end of VRR range.
Can you try overclocking to 76Hz-78Hz, and raise the lower end to 38.5-39Hz? See if that solves the problem....you're getting the underclock artifact at the bottom end of your VRR range, maybe you won't have an overclock artifact at the top end.
You're in a rather interesting situation where your VRR range is 1.87x (40Hz-75Hz) and only need the remainder 0.23x+ either via underclocking/overclocking the ends of your VRR range.
Fixed-Hz 39Hz aren't a reliable test of VRR 39Hz at 75Hz, because VRR 39Hz is like a Large Vertical Total on 39Hz, at a ratio of approximately 75:39 in blanking interval size (plus minimum VBI size at max Hz). Also, refresh rate slewing is sometimes the cause of the blackouts, not the Hz. (e.g. frametime change suddenly changing from 1/75sec to 1/30sec and vice versa) -- sometimes rapid refresh rate slew patterns causes the flicker artifact. Testing reliability of fixed Hz using VRR timings is done via ToastyX, locking the dotclock or horizontal scanrate then increasing Back Porch until the refresh rate lowers to the desired Hz. Then now you're got a VRR-simulating low-Hz fixed-Hz signal (it's not a standard VESA CVT, VESA CVT-Reduced, or other standard timings formula)
To avoid that, you want to try to narrow your VRR range and try avoiding problematic Hz that causes blackouts, by shifting around your VRR range a little bit until you hit a 2.0x VRR range that doesn't have artifacts.
TL;DR: If VRR range-bottom underclocking fails, try VRR range-top overclocking. Or a combination.
Future: Driver enhancements can theoretically add partial LFC to 1.8x and 1.9x VRR ranges by avoiding the problematic underclock range (= slight extra stutter at certain framerates), and I think the native AMD Radeon FreeSync drivers already do it, but NVIDIA drivers does not (yet). There's a continuum of increasing LFC reliability for wider VRR ranges.