Any issues with running at unlimited FPS?

[turnandbank 247]

Silky smooth here locked at 30.  FPS bounce all over the place and stutters galore with unlimited.

[martin-w 3,289]

I run totally unlimited. No internal limiter and no external limiter.

 

When I started doing this it was an absolute revelation for me.

 

Huge frame rate increase and no stutters.

 

From time to time it pays to test conventional wisdom. Try it, you might be VERY glad you did.

 

The caveat is of course, that I run with very sensible slider settings. It's probably the guys with hyper detained scenery and high slider settings that don't experience the magic of unlimited.

[ZKOKQ 986]

Unlimited for me too. Much smoother.

[TheFamilyMan 286]

Silky smooth here locked at 30.  FPS bounce all over the place and stutters galore with unlimited.

 

This is what I experience too as a DX10 fixer user.  When using unlimited frames my FPS is in the 50's while everything is a stuttering mess.  But I will not claim 100% stutter free using NI 1/2 refresh vsync and 30 internal lock.  Hard banked turns in some locations may produce a stutter or two, but almost all of the time othewise silk prevails.  It's suprising that this behavior is so inconsistent from user to user...thus standard disclaimer applies here in spades: YMMV.

[RickB1293 9]

I don't use DX10 because my two favorite airports, FSDT KLAX and PHNL don't work in that mode. Using windowed mode and frame rate locked at 30 within FSX gives me good results. I tend to rely on subjective performance criteria, ie "look and feel", rather than FRAPS reported frame rates or internal FSX numbers. I do notice quite an improvement with the T7 vs the NGX at busy airports like KJFK or EGLL if I turn off AI traffic. If I keep the AI traffic at a modest setting, say 15-20% then the problem goes away and I can get smoother scenery loads with no lag, no blurries and no tearing. I enjoy flying more than tweaking so I tend to not experiment with settings much. Generally, a good OC and a mid to high end graphics card will handle FSX pretty well.

[martin-w 3,289]

Given that some of us get fantastic results with unlimited, and others get a stuttery mess, there's clearly a variable at work.

 

Hyper detailed scenery add-on's, high slider settings, high AA settings, lower overclock that's my hypothesis.

[FLEX1978 373]

I have pretty much every maxed out except Boat & Aircraft traffic.  I fly on vatsim and keep road traffic at 10%

[Christopher Low 5,433]

This is why telling people what "tweaks" to use is a waste of time. You could have a PC with exactly the same specification as someone else, and yet I could almost guarantee you that FSX would not run in an identical manner. I reckon there's voodoo magic at work in FSX. Nothing else makes sense :wink:

[AirborneZed 168]

I limit them to 60 in NI. I never understood the "30 FPS smooth" thing.  It just doesn't look smooth to me. Maybe it does in fullscreen, but not in windowed.

 

The main problem with locking to 30 is that in detailed areas, you'll get 15.   When I lock them to 60, I get 25-30 in detailed areas.  Even more if I set them to unlimited, but then I get tears so I keep it at 60.

[Christopher Low 5,433]

 

The main problem with locking to 30 is that in detailed areas, you'll get 15.

 

Only if you have Vsync set to "ON".

[MDF86 123]

I limit them to 60 in NI. I never understood the "30 FPS smooth" thing.  It just doesn't look smooth to me. Maybe it does in fullscreen, but not in windowed.

 

The main problem with locking to 30 is that in detailed areas, you'll get 15.   When I lock them to 60, I get 25-30 in detailed areas.  Even more if I set them to unlimited, but then I get tears so I keep it at 60.

 

I know what you mean, trying to get FSX right reduces me to tears too :lol:

[SteveW 1,792]

Sat in the cockpit, paused, looking at the sky in front, the unlimited fps readout shows jumping around up to 60fps, but now I set at 60fps limited the fps readout steadies a bit less jumpy but now only at around 50fps. Work has been done and a slight improvement in stability has been gained, even though the fps chosen cannot be maintained.

 

The apparently weird behaviour with the fixed fps setting is because of the way DX makes use of prepared frames and always does some work. Since most often the next frame is largely the same as the last, caching improves performance for work done on prepared frames. Pre-rendering is only completed if there is an abundance of time in the game loop, otherwise the frame is completed to present as the next. There’s more going on under the hood than simple fps capping with the internal limiter. We lose ultimate fps, but gain stability, it’s what it’s designed to do (don’t flame me because this does not happen on your rig).

 

In unlimited mode rendering on the next frame is started as soon as the last is completed. I suggested in an earlier post with unlimited fps to take the outside view and look around in a spin, we notice the time to draw each frame is different. The angular velocity of our view changes constantly with the direction of view, to accommodate changing draw times. Continued panning, speeds up and slows down as the view turns.

 

When we use a fixed fps that can be sustained during panning the view, the angular velocity remains constant. The turning view appears to become unstuck, or asynchronous, from the constraints of the simulation speed. A fixed fps setup allows the sim to try to display frames with an equal time period between each, so that aircraft movement, especially banking and turning, can be more accurately rendered.

 

Generally it does seem that identical machines produce different results, but I would usually put that down to changing circumstances during testing. To make a proper comparison of tweaks and modifications, a method is required that ensures the sim starts with the same parameters each time. For example a .FLT file can be used to ensure the same starting location, time, and view.

[MDF86 123]

Sat in the cockpit, paused, looking at the sky in front, the unlimited fps readout shows jumping around up to 60fps, but now I set at 60fps limited the fps readout steadies a bit less jumpy but now only at around 50fps. Work has been done and a slight improvement in stability has been gained, even though the fps chosen cannot be maintained.

 

The apparently weird behaviour with the fixed fps setting is because of the way DX makes use of prepared frames and always does some work. Since most often the next frame is largely the same as the last, caching improves performance for work done on prepared frames. Pre-rendering is only completed if there is an abundance of time in the game loop, otherwise the frame is completed to present as the next. There’s more going on under the hood than simple fps capping with the internal limiter. We lose ultimate fps, but gain stability, it’s what it’s designed to do (don’t flame me because this does not happen on your rig).

 

In unlimited mode rendering on the next frame is started as soon as the last is completed. I suggested in an earlier post with unlimited fps to take the outside view and look around in a spin, we notice the time to draw each frame is different. The angular velocity of our view changes constantly with the direction of view, to accommodate changing draw times. Continued panning, speeds up and slows down as the view turns.

 

When we use a fixed fps that can be sustained during panning the view, the angular velocity remains constant. The turning view appears to become unstuck, or asynchronous, from the constraints of the simulation speed. A fixed fps setup allows the sim to try to display frames with an equal time period between each, so that aircraft movement, especially banking and turning, can be more accurately rendered.

 

Generally it does seem that identical machines produce different results, but I would usually put that down to changing circumstances during testing. To make a proper comparison of tweaks and modifications, a method is required that ensures the sim starts with the same parameters each time. For example a .FLT file can be used to ensure the same starting location, time, and view.

 

Thanks for the detailed and interesting explanation.

 

However this suggests that a limiter should increase smoothness even if that if at expense of the quoted "FPS" you are getting, whereas I've found through various different tests (all on the same saved approach) that the best smoothness I achieve is most definitely through the unlimited FPS. I've tried internal limiting, external, vSync on and off in NI, various different setups, but I keep coming back to the unlimited FPS in both sim and NI as that's what I find gives me the smoothest flight into a busy detailed airport, in a complex airliner (which is the only way I fly pretty much).

 

So many variables and so little consistency, even between people with almost identical setups! Is it any wonder we all end up chasing our tails in pursuit of the "ultimate" setup?!

[SteveW 1,792]

In your case it seems perfectly fair that going for ultimate fps is what counts in those circumstances.

[Capt Speirs 52]

I don't use DX10 because my two favorite airports, FSDT KLAX 

 

Hmmm, I am not having any issues with FSDT KLAX in DX10.