Frametime fluctuations when v-synced in PC games?

[EchoDissolve]

Hi there. I've noticed that my frametimes in most games fluctuate quite a lot, and this is especially noticeable compared to the console videos I see on Digital Foundry. For example, a stable vsynced 60fps console title on their graphs will be locked at a 16.7ms frame time. I can play the same game on PC, and while the frame rate stays solid, I notice the frametime fluctuates by a small amount, e.g. 16.4 / 17.1 / 16.6 / 16.2 etc. I'm sure the average would be 16.7 over time, but I wonder if this means my game actually isn't as smooth as it could be?

I tried locking to 60fps with RTSS and that actually solves this problem nicely, but I'd still like to know why these fluctuations occur. I've been playing Far Cry New Dawn recently at 4k/30fps using the in-game cap (rather than RTSS) and I see the same problem -- frametimes don't lock to 33.33, but rather bounce around a small window: 32.8 / 33.1 / 34.1 / 33.7 etc. Does this mean my game at 30fps on PC isn't as smooth as it would be on console at 30fps? That doesn't sound right to me, but the numbers are there.

I know it's just my neurotic side coming out, but is there a reason why many PC games don't have locked frametimes like console games do? And is locking with RTSS really a good solution to this, or do these fluctuating numbers not actually introduce any stutter when v-synced? I don't feel any stutter, to be honest, but the numbers make me wonder. Is there a difference in smoothness between in-game capping and RTSS?

Thanks in advance!

[jorimt]

A lot can be said on this subject, and yes, what you're seeing is normal, but to start, I did post something related to this in an answer to another user's inquiry a while back:

viewtopic.php?f=10&t=4610&start=10#p35826

"Frametime" reflects render time, and render time can, at points, be "faster" than the max refresh rate with the framerate above it with V-SYNC ON. This is why you don't see the FPS meter on, say, Afterburner, go over your max refresh rate with V-SYNC ON, but why you can still see the frametime meter sometimes dip (lower frametime = higher framerate) when the framerate exceeds your refresh rate with V-SYNC ON.

Lower frametimes (which, again, reflect "faster than refresh rate" render times) is what causes the over-queuing of V-SYNC buffers without a proper FPS limit in place, which, in turn, ultimately causes, yes, the repeated render "blocking," and thus added input lag with V-SYNC ON.

And when I said "V-SYNC does not stop the system from rendering more frames" in my last post (again, obviously too much oversimplification in trying to explain it to you on my part), I meant it doesn't stop it in time to prevent the added delivery delay when the framerate is well above the refresh rate without an FPS limit below the refresh rate in place.

So short answer, no, the frametime variances you are seeing won't negatively affect framepacing with standalone double buffer V-SYNC with framerates sustained above your refresh rate.

If input lag is a concern to you, however, with standalone double buffer V-SYNC + framerate sustained above your max refresh rate (say 60 FPS @60Hz, or 144 FPS @144Hz), you typically want an RTSS limit slightly below your refresh rate to decrease V-SYNC input lag:
https://www.blurbusters.com/howto-low-lag-vsync-on/

For "console-like" 30 FPS on PC with standalone double buffer V-SYNC, you typically need to set the game (on a 60Hz monitor) to 1/2 refresh rate V-SYNC via Nvidia Inspector (Nvidia GPUs only; I'm not aware of an AMD equivalent) AND set a 30 FPS limit via RTSS.

[RealNC]

Digital Foundry measures frame times using video capture. Of course it's going to show up as rock-solid, since a video signal is never going to fluctuate. It's fixed 16.7ms. If the game drops in framerate, you get jumps between 16.7ms and 33.3ms, but never a value in-between.

If you would measure the same way on a PC, using a non-VRR display, you'd get the same results; rock-solid 16.7ms, rock-solid 33.3ms, or constant jumps between 33.3 and 16.7.

When you use RTSS to measure, you're not measuring what you're seeing. You're only measuring what the game is sending to the GPU, which is NOT what you're seeing. What you're seeing is what the GPU is sending to the monitor, and to measure that, you'd need to measure using the video capture method Digital Foundry uses.

[EchoDissolve]

Thanks both for your quick replies! Seems like there's a lot to dig into on this topic, so down the rabbit hole I go!

[EchoDissolve]

For "console-like" 30 FPS on PC with standalone double buffer V-SYNC, you typically need to set the game (on a 60Hz monitor) to 1/2 refresh rate V-SYNC via Nvidia Inspector (Nvidia GPUs only; I'm not aware of an AMD equivalent) AND set a 30 FPS limit via RTSS.

Would setting maximum pre-rendered frames to 1 globally have the same effect as enabling the 30 fps RTSS lock?

Additionally, if a game has the in-game option to change screen refresh rate to 30hz and enable vsync, does this achieve the same as enabling half-refresh vsync?

[jorimt]

Would setting maximum pre-rendered frames to 1 globally have the same effect as enabling the 30 fps RTSS lock?

No, the first can lower input lag noticeably at lower FPS (assuming the given game respects the MPRF override), and the second ensures proper frame pacing.

Additionally, if a game has the in-game option to change screen refresh rate to 30hz and enable vsync, does this achieve the same as enabling half-refresh vsync?

Theoretically? Possibly, but I've never tested it directly myself (I've always had Nvidia GPUs when I've used this method for a 30 FPS lock, and now that I have G-SYNC, I no longer need to rely on standalone V-SYNC methods in these instances).

[EchoDissolve]

Thanks for your help. I had some issues with tearing at half-refresh with the RTSS lock, but I disabled the adaptive part of the half-refresh and I think I now have a smooth "console 30fps" for Far Cry ND at 4k on a 1070ti. Thank you!

One last thing I'm curious about: What's the need for the RTSS lock on top of the adaptive half-refresh? Without the lock I saw frame times locked to 33.33ms, but input lag did seem worse than with the lock. Happy to hear a technical explanation if you have one here -- always like to learn.

Thanks again!

[jorimt]

The Nvidia control panel only has the adaptive 1/2 refresh rate V-SYNC option exposed. You need to download Nvidia Inspector and use the non-adaptive 1/2 refresh option there:
https://ci.appveyor.com/project/Orbmu2k ... /artifacts

That will solve the tearing.

[EchoDissolve]

Thanks for your help. I had some issues with tearing at half-refresh with the RTSS lock, but I disabled the adaptive part of the half-refresh and I think I now have a smooth "console 30fps" for Far Cry ND at 4k on a 1070ti. Thank you!

One last thing I'm curious about: What's the need for the RTSS lock on top of the half-refresh? Without the lock I saw frame times locked to 33.33ms, but input lag did seem worse than with the lock. Happy to hear a technical explanation if you have one here -- always like to learn.

Thanks again!

[Notty_PT]

In my opinion no matter what you do on PC, if you play a 30fps game with a mouse it will always feel bad. It feels better with a controller. Still, on console everytime you move the camera I notice the lack of motion clarity due to the low framerate.

Samsung TVs have an interpolation technique called game motion plus wich is great and smooths the image to PC 120hz standards, altho with a bit added input lag (21ms).