Page 1 of 1
Has anyone tested Fast-Sync vs Fullscreen Borderless with optimizations for windowed games?
Posted: 01 Jun 2024, 03:07
by Gidrah
Has anyone tested the input lag difference between Nvidia Fast VSync and Fullscreen Borderless with Optimizations for windowed Games enabled (flipped presentation model) on Windows 11? I've search for a while and haven't found anything solid.
I am trying to apply this to games like Cyberpunk and Elden Ring where I set a 60fps cap in RTSS but get terrible screen tearing with V-Sync off on a 165hz monitor. Unfortunately my laptop is not a VRR monitor so unless there's a better option I must choose between the above for a tear free low latency experience. This may even be a situation where NULL would be helpful but I haven't researched it.
Side question. I was curious if HDR has any effect on input latency. There's conflicting information online but this is specifically for my home monitor the MSI MPG 271QRX QD-OLED. I am VERY NEW to HDR but so far it looks better on so I wanna make sure I am not trading anything.
Thank you for any help!
Re: Has anyone tested Fast-Sync vs Fullscreen Borderless with optimizations for windowed games?
Posted: 07 Jun 2024, 03:53
by Gias
the difference between fast sync and fullscreen borderless with optimizations for windowed games on windows 11 ? what do you mean ? I mean they're kinda for different purposes and they can work together... in applicable render APIs that is.
keep in mind that windows 11's optimizations for windowed games may force flip model for windowed games but it's only for d3d10/d3d11 games, and windows/gamebar needs to recognize the game as a game (otherwise the optimizations for windowed games won't be applied), so for some games (typically new ones...) you may need to check the "remember this is a game" option in the gamebar settings for the game. also, microsoft has an internal list where fullscreen optimizations and optimizations for windowed games may be disabled for some games. though i believe FSO is enabled for most...
and fast sync does not work in d3d12 games, so makes sense that you were getting tearing in cyberpunk and elden ring with vsync off or fast sync (both cyberpunk and elden ring use d3d12).
fast sync may work with d3d9ex and d3d10/d3d11. it may also work with vulkan or opengl games that are using d3d11 for final presentation (such as with nvidia's dxgi swapchain interop for vulkan/opengl, or special k's opengl-ik -- which uses d3d11 for final presentation)
anyway, for a lower latency sync (at least compared to vsync on) approach with d3d12 (though not limited to d3d12), you could try vsync off with rtss's scanline sync or special k's latent sync. with either one of those, you could move and hide the tearline so that it's not visible to you. though the effectiveness may depend on gpu load or your hardware and the game.
you can also try rtss's scanline sync in hybrid move with rtss's back edge sync to prioritize lower latency or rtss's scanline sync in hybrid mode with rtss's front edge sync to prioritize tearline position stability.
similarly, special k's latent sync also has options to prioritize lower latency or tearline position stability thorough latent sync's delay bias settings. sk's latent sync delay bias by default uses the "no input, all display" mode, which basically behaves like rtss's scanline sync in hybrid mode with front edge sync, and then latent sync's delay bias has various percentages you could set towards input to lower latency at various steps (the higher input delay bias, the lower the latency; however, how much you can push or increase latent sync's input delay bias for lower latency without or before getting pacing issues and/or syncing issues basically depends on your hardware and the game...)
special k's latent sync also has an "allow tearing" option that is checked by default. if for some reason you're not able to reliably keep the tearline out of view with latent sync's "sync offset" slider/options... then you could uncheck latent sync's "allow tearing" option to stop the tearing in d3d11 and d3d12 games.
what that "allow tearing" setting does is control the dxgi present tearing flag for windowed games (all d3d12 games technically use a windowed mode even in fullscreen. in fullscreen it's technically windowed borderless fullscreen, and d3d12 always uses flip model swapchains). the operating system, windows, by default does not allow tearing for windowed/borderless apps, but a way to make it possible is with the dxgi tearing flag (and dxgi games that use flip model normally use that to allow tearing). technically then when you're unchecking special k's "allow tearing" box, what you're also doing is essentially bringing back the "OS's vsync" for windowed/borderless apps (the windows desktop manager's composition is still bypassed by the game with flip model if directflip/independent flip optimizations have been engaged, but you'd basically have a form of vsync on enabled by the operating system if sk's "allow tearing" box is unchecked), which may result in higher latency; however, latent sync's delay bias settings still work, so you could still increase the input delay bias percentage to reduce latency some (even with the "allow tearing" box unchecked).
i believe some people also use rtss's scaline sync + fast sync or sk's latent sync + fast sync, which could work well in certain cases. fast sync still won't work in d3d12 keep in mind.
Re: Has anyone tested Fast-Sync vs Fullscreen Borderless with optimizations for windowed games?
Posted: 28 Jun 2024, 02:36
by Gidrah
Thanks for the reply. I've just been using borderless for elden ring and cyberpunk as it seems to feel the most responsive without tearing. I had tried rtss scanline sync for Elden Ring once and was not able to get it to work properly due to that game being locked to 60fps. I can also tell that game is doing weird things to the refresh rate as my monitor reports 120hz in fullscreen and the full refresh rate which feels better anyways. I decided to stop worrying about it and just go by what "feels" better.
My new home monitor is the AW2725df and supports G-Sync and I may end up running that with Elden Ring in full screen to take advantage of HDR but I have to test if I prefer that to Wide Gamut or sRGB.
Re: Has anyone tested Fast-Sync vs Fullscreen Borderless with optimizations for windowed games?
Posted: 01 Jul 2024, 22:29
by Gias
Gidrah wrote: ↑28 Jun 2024, 02:36
Thanks for the reply. I've just been using borderless for elden ring and cyberpunk as it seems to feel the most responsive without tearing. I had tried rtss scanline sync for Elden Ring once and was not able to get it to work properly due to that game being locked to 60fps. I can also tell that game is doing weird things to the refresh rate as my monitor reports 120hz in fullscreen and the full refresh rate which feels better anyways. I decided to stop worrying about it and just go by what "feels" better.
My new home monitor is the AW2725df and supports G-Sync and I may end up running that with Elden Ring in full screen to take advantage of HDR but I have to test if I prefer that to Wide Gamut or sRGB.
yeah elden ring is normally locked to 60fps. you can use Special K to unlock elden ring's fps (but you'd be forced to play offline)
you can also use elden ring's native HDR in a proper windowed/borderless mode thanks to SK's fake fullscreen option (as long as you're using an nvidia gpu since SK's fake fullcreen option currently works with nvidia gpus only)
https://wiki.special-k.info/en/SpecialK ... Elden_Ring
or use SK's own HDR Retrofit feature (mostly for games that don't have a native HDR option, but I believe some people still prefer SK's HDR retrofit over the native HDR of some games)
https://wiki.special-k.info/HDR/Retrofit