Vsync and frame limit hurt performance

[cpm9]

I know vsync caps fps to your monitors refresh rate, and I know the frame limit caps to the limit you set. However, when comparing performance it seems that the vsync and frame limit hurt performance.

If fps is always above 200 with vsync off and frame limit unlimited but you turn on vsync on a 144hz monitor it will go to 144 fps but half the time it will be between 100-144 fps. And the same for a 144 frame limit. So compared to the always above 200 on vsync off unlimited the performance is lesser.

Maybe something in the vsync and frame limit code could be improved to fixx this? I hope my explanation makes sense. Can anyone replicate this problem?

magnesium-1.4.1.1.jar

[Quinteger]

That's how vsync and frame limiting work. They are essentially just timers that fire based on how much time did it take to render last frames. If you got a difficult frame, you will get a performance dip.
Goes without saying that it's not related to sodium.

[cpm9]

That's how vsync and frame limiting work. They are essentially just timers that fire based on how much time did it take to render last frames. If you got a difficult frame, you will get a performance dip. Goes without saying that it's not related to sodium.

So you're telling me that when I am getting above 200fps then I turn on vsync or frame limit to 144 but it goes to 100 that is normal? Why does vsync/frame caps not work like this in other games?

[Quinteger]

That's how vsync and frame limiting work. They are essentially just timers that fire based on how much time did it take to render last frames. If you got a difficult frame, you will get a performance dip. Goes without saying that it's not related to sodium.

So you're telling me that when I am getting above 200fps then I turn on vsync or frame limit to 144 but it goes to 100 that is normal? Why does vsync/frame caps not work like this in other games?

Yes.
FPS is not a metric you can calculate instantly. You need some time period to calculate it against and different games can pick different periods.
If on a 60 fps vsync you got a difficult frame which took more than 1/60s, but less than 2/60s to render, you can say that your fps during short period was 30, but a different game that measures FPS over a 100ms period will report to you something like 50 for example.

[cpm9]

Okay I have done more investigation and found what I think is good proof of what I've reported. In these screenshots I've used the lowest settings and 2 render distance to try to best demopnstrate the problem. Tested with Magnesium 1.5

@Quinteger

Number one: vsync off and frame rate unlimited. Average fps is ~1600 and 1% low and 0.1% low are ~700

Number two: vsync on frame rate unlimited. Average fps is 144 (like monitor refresh) but 1% low is 117 and 0.1% low is 43 this microstutter makes it feel a lot less smooth and was not present without vsync. There is also the fact that the current fps drops below 144 to 143 or lower every few seconds. I don't understand how this makes sense considering the huge fps numbers when unlimited and the huge 1% and 0.1% lows.

Number three: vsync off and frame rate set to 145. Same as above.

[cpm9]

[cpm9]

I think this is a limitation of minecraft itself and not specific to Magnesium