Vsync setting and impact on performance [update: Nvidia driver 517.48 vs 516.94]

User Support Hub Install, Performance & Graphics
, ,
1

[9/27/22 update]: added performance comparison between Nvidia drivers 517.48 and 516.94

I wanted to start some benchmarking with SU10 to determine the settings that would make the best use of my hardware. I started by looking into Vsync as there does not appear to be any consensus on whether Vsync should be enabled in the game settings or in the Nvidia control panel.

  • Testing scenario: All measurements were performed during the first 20 seconds of the NYC discovery flight. All tests conducted using DX11. Performance metrics consist of: FPS, FPS stability (measured in terms of time differences between two consecutive frametime values), average CPU load, and average GPU load.
  • My specs are: i7-12700KF, RTX 3080, 32GB DDR5, 1TB NVMe, 3440x1440 (ultrawide) 60Hz monitor.

1- Effects of Vsync mode on fps

  • As expected: uncapped Vsync modes result in the highest fps; capped modes (Nvidia adaptive at 50% refresh rate and MSFS Vsync at 50% refresh rate) lock at 30 fps
  • No noticeable difference between the two Nvidia driver versions

fps
fps4167Ă—1724 260 KB

2- Effects of Vsync mode on fps stability
Things get more interesting when looking at how stable the fps are under each Vsync scenario.

  • The “fast” and “adaptive” Vsync options available in the Nvidia control panel result in the most stable states (less than 4 ms between consecutive frames for nearly 98% of all frames captured).
  • The least stable fps is obtained when Vsync is set to “on” in the Nvidia control panel.
  • Regardless of the driver version, Nvidia Vsync fast and adaptive yield the highest degree of frame rate stability.

variance
variance4167Ă—1723 257 KB

3- Effect of Vsync mode on CPU/GPU load

  • As expected, both fps capping options substantially reduce the load on the CPU and GPU, while uncapped options result in similar CPU/GPU loads.
  • The increase in frame rate stability seen with Nvidia Vsync fast does not adversely affect CPU and GPU loads as compared to the other uncapped Vsync modes.

CPU and GPU load
CPU and GPU load4167Ă—1720 185 KB

Conclusions:

  • Based on the data collected, it looks like Vsync fast in the Nvidia control panel is able to maintain a high frame rate in a more consistent way than any other Vsync modes. Surprisingly, I have not seen any post suggesting the implementation of this mode in MSFS.
  • Capping the frame rate does not seem to increase frame rate stability (both capping modes actually yielded the least amount of stability in my tests).
15 Likes
2

Thanks for posting this - this is a nice guide for people to understand how Vsync affects performance.

It might be worth noting these tests are done in DX11. For DX12, Fast Vsync is not an option in the Nvidia Control Panel due to how DX12 works.

1 Like
3

I still don’t understand why the two options capped at 30 fps result in such a noticeable deterioration in frame rate stability. I would have thought that relaxing the load on the GPU would automatically ensure less variability.

4

This is a great report.
What do you think is the most stable and good setting?
I limit the frames to 40 and play.

It helped a lot. Thank you.

5

Visually, fast Vsync provided the most “fluid” experience, so I think I’ll try this for a while. But again, I would expect that capping the frame rate would help with fps stability. Maybe I need to test again with a different scenario, e.g., dense and active airport.

6

Benchmark now includes a comparison between Nvidia drivers 517.48 and 516.94

7

Thank you for this. Really great work!

I’m not sure what’s wrong with my setup, but the only way for me to stop tearing is to use the in game vsync. I’m glad to see there’s hope though that someday I’ll be able to use different and better settings.

1 Like
8

Just a noob question:
If I set “Vsync Fast” in Nvidia Control Panel, should I enable the ingame vsync?

9

If you set it to fast in the Nvidia control panel, you should turn Vsync off in the game settings.
I need to do more testing though with Vsync fast vs. capping frame rate and no Vsync. I did a test flight yesterday night with the FBW A32nx, taking off from LFMN and I was not impressed. A lot of stuttering. I want to see how capping the frame rate and using no Vsync can unload the CPU and GPU.

1 Like
10

So, I ran more tests yesterday and my conclusion is that benchmarking MSFS based on fps, fps variance or CPU/GPU load alone is somewhat irrelevant as the perception of the flying experience differs substantially from the expected experience as quantified by those metrics.
First, shortly after updating to SU10, I experienced noticeable performance deterioration with frequent stutters and fps jumps. This was especially noticeable with the FBW A32nx and heavy airport sceneries (e.g., LFMN). I completely uninstalled and reinstalled the game and the experience was much improved.
Even with the improved performance following reinstallation, setting Vsync to “fast” in the NVidia control panel did not yield the best flying experience. After testing many scenerios, I settled on leaving all Nvidia control panel settings at default, and implementing Vsync in the game at 50% refresh rate. While fps capping will disturb some, for me it eliminated nearly all micro-stutters while taxiing in large airports, without noticeably affecting “smoothness”. This was by far the setting that provided the best flying experience.
The bottom line is that, for this game, metrics do not tell the whole story. The experience behind the yoke matters more!

1 Like
11

Completely agree with your assessment. I found the same results with experimenting with various settings. Thanks for publishing your test findings.

1 Like
12

As I understand it, which is probably incorrect, v-sync tries to match your monitor’s refresh rate.

Usually with v-sync on (game settings) I get an average of 60 fps in low populated areas. With v-sync off I get an average of 70 fps, but I get screen tearing on panning.

My GPU is an RTX 2060. My monitor is an ASUS VG248QE 24” 144hz. I have no idea if that monitor is considered good, fair or dog, but I’m happy with it.

If I set my game settings to 50% refresh rate, does that mean it will effectively aim to reach a level of half the monitor’s refresh rate?

13

Yes it will. My monitor is 60 Hz and if I enable Vsync at 50% refresh rate in game, I get a steady 30 fps.

1 Like
14

Thank you :+1:

15

Is the in game vsync and frame limiters tied to the VR headset, eg my G2s 90hz or my TV, ie 120hz. Would the limiter at 50% be limiting to 45 or 60fps? I dont know how vsync timing works with both vr and tv displaying images at the same time.

16

The guys who are getting screen tearing - do you have a gsync monitor with gsync turned on?

Also of note is the way gsync works together with different vsync modes, which can explain why some get different results especially in terms of perception vs benchmark.

There’s some research that says regular vsync ON is best if you also have gsync instead of using fast/adaptive. It’s a complex topic though if you want to dive down that rabbit hole G-SYNC 101: Optimal G-SYNC Settings & Conclusion | Blur Busters

QUOTE from the site about fast vsync, bolded sections in the last paragraph is what may be relevant to the perception of less stability and stutters:

Fast Sync only begins to reduce input lag over FPS-limited double buffer V-SYNC when the framerate far exceeds the display’s refresh rate. Like G-SYNC and V-SYNC, it is limited to completing a single frame scan per scanout to prevent tearing, and as the 60Hz scenarios show, 300 FPS Fast Sync at 60Hz (5x ratio) is as low latency as G-SYNC is with a 58 FPS limit at 60Hz.

However, the less excess frames are available for the third buffer to sample from, the more the latency levels of Fast Sync begin to resemble double buffer V-SYNC with an FPS Limit. And if the third buffer is completely starved, as evident in the Fast Sync + FPS limit scenarios, it effectively reverts to FPS-limited V-SYNC latency, with an additional 1/2 to 1 frame of delay.

Unlike double buffer V-SYNC, however, Fast Sync won’t lock the framerate to half the maximum refresh rate if it falls below it, but like double buffer V-SYNC, Fast Sync will periodically repeat frames if the FPS is limited below the refresh rate, causing stutter. As such, an FPS limit below the refresh rate should be avoided when possible, and Fast Sync is best used when the framerate can exceed the refresh rate by at least 2x, 3x, or ideally, 5x times.

Also there’s also low latency mode in Nvidia panel that pairs well, it specifically mentions running it with Gsync + Vsync together:
image

17

Is the synopsis here explaining why I see worse FPS when I cap it at 50% for 30 FPS — the sim struggles to maintain that 30 dipping down into the mid 20s?

When I “cap”’ it at 100%, I see high 30s to low 40s and they remain that way consistently. This in the same test scenario.

In short, “capping” it at 100% actually improves my overall FPS and offers higher performance overall.

18

Lots of good info here. I recommend going through the docs to understand the technology. I remember reading some of the articles here before I bought my system. The info convinced me to spend a few $$$ more to buy a g-sync monitor. The conclusion states:

But if set up properly, and run on a capable system, G-SYNC is the best, most flexible syncing solution available on Nvidia hardware, with no peer (V-SYNC OFF among them) in the sheer consistency of its frame delivery.

19

For me, it’s all about the plane and the scenery. I flew near Gibraltar with the Kodiak and the FBW A32nx. With the Kodiak, I can set my Vsync at 100% refresh rate and I can push frames and have a stutterless flight. With the jetliner, I can achieve the same stutterless experience only if I limit my Vsync at 50% refresh rate, due to the increased CPU load imposed by this plane.

1 Like