How do you benchmark your settings in MSFS and NCP? I mean, how do you know, whether the changes you just made are improving your fps and consistency of time-frames?

I ,of course, know the fps counter in DevMode and I use the benchmarking tool embedded in the MSI Afterburner measuring average fps over a period of time and corresponding  fps of 1% and 0.1%. But I am looking more for an approach in which situation to measure this and how to ensure that I am measuring always the identical period of flight and in identical conditions.

I tried to set up always the same flight or save the flight. Then to fly it on an autopilot and start measuring always when a certain building passes under the windshield edge. Then to measure for a period of 2, 3 or 5 or even 10 minutes. Despite switching off all weather (esp. clouds constantly change and if I fly through them have big impact on avg. fps) I was unable to get consistent readings let alone to base the effect of small variations in settings on them.

The ideal solution would be some fully automated benchmark like 3DMark Time Spy.. I know that some websites used the Sydney landing challenge as a benchmark, but it is mystery to me how they ensured to fly exactly the same path every time.

And yes, you could say that the effect of various settings in MSFS and in NCP is highly individual and that it is the best just to fly, not to look at fps, and if it feels good then it is good. Yes, I agree, but I honestly cannot tell the difference between 55, 60 or 65 fps. What if some of the settings I deliberately use e.g. in NCP based on various "recommended" guides are in fact costing me 3 or 5 fps? There are so many recommendations around and we all know that each system is in fact unique.

How do you please benchmark your settings?

"I honestly cannot tell the difference between 55, 60 or 65 fps...What if some of the settings...are in fact costing me 3 or 5 fps?" You first indicate you can't tell the difference between 55-65 fps (so I assume that's what you're used to in MSFS) but are then worried you may be cheating yourself of 3-5 fps by possibly using less than optimal settings? Not sure I understand the problem?

I cant tell the difference between 23fps and 60fps.😨

I use CapFrameX for VR benchmarking, works very well for my use.

5 hours ago, MarcG said:

I use CapFrameX for VR benchmarking, works very well for my use.

CapFrameX definitely looks very promising with all its statistics. I did not know this tool, thank you very much, MarcG!

Now I just have to figure out how to be as consistent as possible between individual benchmarking flights.

8 hours ago, somiller said:

You first indicate you can't tell the difference between 55-65 fps (so I assume that's what you're used to in MSFS) but are then worried you may be cheating yourself of 3-5 fps by possibly using less than optimal settings? Not sure I understand the problem?

Yes, that's exactly the situation. I guess I did not explain it properly in the initial post, sorry.

I have max. frame rate set to 60 fps in NCP. While flying the frames stay in 95-98% of cases at 60. In very bad moments they drop to 35-40. But it is of course the 2-5% you are aiming to eliminate as it destroys the fluidity. Plus there are occasional microstuttters, for example while panning.

The goal simply is to tune the settings for maximum performance possible (+ keeping reasonable quality) using the given hardware. And since there are dozens of settings (e.g. Negative LOD bias or Low latency mode) , some benchmarking seems to be necessary.

It is the same situation as with a car. When driving you are satisfied with 80% of the engine performance, because you don't need the rest. But when you are overtaking you are glad to have the extra power.

9 hours ago, MarcG said:

I use CapFrameX for VR benchmarking, works very well for my use.

Dear MarcG,

Just read your post from January 21, 2021:

https://www.avsim.com/forums/topic/593383-vr-benchmarks-for-nvidia-drivers/page/2/

It describes in high detail exactly what I was looking for: How to measure multiple times exactly the same flight. Thank you very much for that!

May I ask: What do you mean by Toolbar AP, please? 

12 hours ago, MarcG said:

I use CapFrameX for VR benchmarking, works very well for my use.

> May I ask: What do you mean by Toolbar AP, please?

Oh, sorry, I know now. It's the MSFS Toolbar AI Autopilot.

Coming from FSX I dd not know the AI AP is able not only to fly to a certain point, but is able also to take off and land... I have just tried it and the whole benchmarking routine works flawlessly.

Please forget the question.