btarunrNVIDIA explains that it changed this recommendation as the GeForce drivers since 461.09 support MPO (multiple plane overlay). Prior to 461.09, the Windows 10 DWM (desktop windows manager) would composit the contents of all windows into a single plane and presented at display refresh-rate cadence, so tearing was avoided even with V-Sync globally off. With MPO, applications are allowed to present their contents independently, each with their own V-Sync setting. The OS, outside of driver control, decides which apps get promoted to their own MPO plane. This would lead to screen-tearing if V-Sync is globally off, even in Windowed or Windowed-Maximized (pseudo-fullscreen) modes.

So is this Nvidia fixing Microsoft's problem or am I misunderstanding the changes?

They really should re-design their User Interface at this point. Especially when you re-select options, it keeps jumping the list upward...

Legacy-ZAThey really should re-design their User Interface at this point. Especially when you re-select options, it keeps jumping the list upward...

Please no. They'll just make something worse that takes 10GB of hard drive space and runs like a snail while simultaneously pegging one core to 100% at all times.

ZeppMan217So is this Nvidia fixing Microsoft's problem or am I misunderstanding the changes?

It seems to me that it was meant to be like this. Though this would have taken a long time, since MPO was introduced way back with Windows 8.1.

ZeppMan217So is this Nvidia fixing Microsoft's problem or am I misunderstanding the changes?

It depends on your definition for "problem." Infrastructure changes do often require the superstructure to accommodate them, that's unavoidable. The argument then becomes on whether the gains outweigh the costs.

That said, this change wasn't -afaik- mandatory to adopt. It did take nvidia years to do so. So it was partially Nvidia's choice to make that change...

hatPlease no. They'll just make something worse that takes 10GB of hard drive space and runs like a snail while simultaneously pegging one core to 100% at all times.

What is an Electron app?

ShihabyoooIt depends on your definition for "problem." Infrastructure changes do often require the superstructure to accommodate them, that's unavoidable. The argument then becomes on whether the gains outweigh the costs.

That said, this change wasn't -afaik- mandatory to adopt. It did take nvidia years to do so. So it was partially Nvidia's choice to make that change...


What is an Electron app?

If discord is any indication, an electron app is precisely what hat was describing :v

I have an Asus Strix 3090 on my PC connected to a LG CX 48" 4K tv/monitor via HDMI 2.1 at 120Hz with HDR and Gsync on.

Things were running rock solid until I installed 461.09 AND a pending Windows update. Big mistake doing both things at once, as I started having the Windows desktop randomly flicker on me. Of course, I blamed the nVidia drivers and was almost about to install 461.40 when I read about the problems that driver version had with LG CX monitors and Gsync, so I didn't.

Then patch Tuesday came with a ton of security fixes for Windows. Well, guess what, despite 'random flicker' not being listed anywhere as something MS had fixed in the latest updates, the fact is that the random flicker is now gone (and I am still running 461.09)

At least now I know what was causing it.

I'm perplexed by how this operates now relative to prior and Microsoft's intentions for it. Can you disable vsync for the GDI now? Prior vsync was always active in windowed mode. Does this mean each display is segmented and can control resolution and refresh rate and vsync on or off? Does that mean the CPU/GPU overhead can actually be decreased or increased effectively over how it was prior? What I mean by that if it was all scared prior in one plane represents all three displays by DWM the overhead is different by not being dynamic in nature. It would seem you could now set left and right side displays to something like 24Hz/30Hz refresh rates and conserved power and overhead in the process with little or no practical side effects. I would think for games if the middle refresh rate were higher you wouldn't experience input lag on those while the mouse pointer is fixed on those regions and normally it would be in practice. Are there any mGPU implications from this as well. I think it's no secret chiplet mGPU designs are being actively pursued by AMD and Nvidia. If anything else from the looks of it on the bottom diagram it makes it more parallel friendly in nature and certainly presents a upside.

460.79 here and it already was marked as recommended.

I'm in 461.40 and I'm afraid. They still have to juggle the compatibility with gsync with this and the different vsync options. Since some time ago some programs just tend to blink and stutter in loading screens and rarely there are some desyncs. For now I have just gsync for everything without any vsync, but gsync and rapid vsync seem to work alright together. Also there are some dpi problems with some apps and I have to force fixed refresh rate on those or everything is blurry in that app.

If umlb didn't darken the screen so much I would be alternating with that, it's a pity.