Measuring performance with Uncapped FPS ?

[kingsal]

Hello, I'm trying to find a good way to measure the performance in my mission.
TLDR: Can uncapped FPS be used as a way to gauge performance on mid / lower end graphics cards?

Typically, I use a combo of FPS, drawcalls, and shadow counts to get some idea of performance. However, these days my machine rarely drops below 60 FPS (when capped). So I started using uncapped FPS to get a rough idea of how heavy a scene is relative to the rest of the mission.

Lets say most of my scenes run at 300 (the hard limit ) and one scene runs at 150 FPS does that roughly mean  a user who normally runs at 60 (capped) will be at 30 in that scene?
I'm not looking for an exact science, just a rough tool to get some idea of performance. 

 

[Wellingtoncrab]

13 minutes ago, kingsal said:

Lets say most of my scenes run at 300 (the hard limit ) and one scene runs at 150 FPS does that roughly mean  a user who normally runs at 60 (capped) will be at 30 in that scene?

Imo no, because the amount of milliseconds it takes to render a frame is very different at low vs high frame rates.

30 FPS is 33.33333 MS
60 FPS is 16.66666 MS
90 FPS is 11.11111 MS
120 FPS is 8.333333 MS
140 FPS is 7.142857 MS
144 FPS is 6.944444 MS
180 FPS is 5.555555 MS
240 FPS is 4.166666 MS

The practical difference between 150 vs 300 is basically 3 ms per frame of render time, vs the over 16ms that comes with 30 vs 60.

[I.C.H.I.]

Install MSI Afterburner and set it to show CPU, GPU utilization both in clock and usage, also enable display of the load on each CPU core and RAM, VRAM consumption by the process. I don't trust FPS because games often have it locked but frametime measurement is much more useful.

[Wellingtoncrab]

An equivalent of 60 to 30  would be something more like dropping from 300 down to like 50-60 frames maybe, but I assume that is still flawed given all the differences in hardware and how they handle the work loads period.

[peter_spy]

21 hours ago, I.C.H.I. said:

Install MSI Afterburner and set it to show CPU, GPU utilization both in clock and usage, also enable display of the load on each CPU core and RAM, VRAM consumption by the process. I don't trust FPS because games often have it locked but frametime measurement is much more useful.

Seconded. I prefer that to the in-game FPS meter, since I can see what impacts the CPU and GPU separately.

[kingsal]

22 hours ago, I.C.H.I. said:

Install MSI Afterburner and set it to show CPU, GPU utilization both in clock and usage, also enable display of the load on each CPU core and RAM, VRAM consumption by the process.

Thank you, Ill give this a try!

 

 

22 hours ago, Wellingtoncrab said:

An equivalent of 60 to 30  would be something more like dropping from 300 down to like 50-60 frames maybe, but I assume that is still flawed given all the differences in hardware and how they handle the work loads period.

Ahh okay this makes a lot of sense. Ill keep that in mind, so its some kind of measurement, just not a very reliable one. Thank you.

[chakkman]

23 hours ago, kingsal said:

Typically, I use a combo of FPS, drawcalls, and shadow counts to get some idea of performance. 

I think you already get a good impression that way. That plus beta testing with different people on different machines (obviously).

I'm wondering how many TDM players are running higher refresh rate displays these days.

Edited May 22 by chakkman

[peter_spy]

Quite recently, I switched to a laptop with 240Hz display. While I don't see that much improvement over 120 Hz/FPS threshold (maybe it's the age and reaction time thing), I do like the improved latency quite a bit. I know it's not that relevant in a stealth game with slower-paced gameplay, but it's just nice in how smooth the game response to input is.

[chakkman]

I switched from a 60 Hz display to a 144 Hz display some time ago. I absolutely love the smoother movement and animations in games. It really makes the games more realistic.

Regarding the input lag: I don't think I would notice that. I don't notice a difference in input lag between Vsync on and off either.

[peter_spy]

I'm a hobbyist musician, mostly playing bass and some basic synths on the side. When you're recording yourself, you can simulate or compensate for the hardware latency in the drivers, and even some small values, like 30-60 ms change how my instruments respond to my playing. Maybe it comes from that