FPS Limited vs. Unlimited

[Kosta 122]

Let's go into a new topic: why is it there, and does everyone have it?Simple thing: VSync off, no FPS limiter, no internal limit: FPS shows for instance 50.You set the limit just to 99 - what happens? FPS drop to 40.Happens with all aircraft and everywhere, no matter what addon you are using. Not tested with vanilla FSX.The drop is usually percentage, I can't say exactly how much, but to me it seems about 25-30%.Can we try and find out why?EDIT: Corrected some things as I did false tests...

[cianpars 1,171]

The framerate thing has always been interesting. I always got smoother performance in FS9 and FSX on mys previous 2 systems limiting to jsut under what I am normally getting, yet on my new system, I get more stutters in FSX if I limit than if I go unlimited.It does seem to be system related in some way. Some users get better performance with and some without. You just have to suck it and see.Some have had success using the external frame rate limiter as that works in a different way. Didn't make any difference for me though.I'm sure someone can apply some logic to it though.IAN

[Bert Pieke 7,585]

Let's go into a new topic: why is it there, and does everyone have it?Simple thing: VSync off, no FPS limiter, no internal limit: FPS shows for instance 50.You set the limit just to 99 - what happens? FPS drop to 40.Can we try and find out why?

Not sure this is the entire explanation, but the texture loading formula is differentfor the unlimited vs the limited setting. I find by looking at CPU utilization ofthe cores, the core that carries the main load will run at a higher utilization whenthe fps are limited within FSX than if set to unlimited. Now, what exactly is keepingthe core busy I do not know... but I hope it is doing something useful which couldbe related to terrain loading of some sort - which might leave less cycles for high frame rates.. If anyone has a better explanation - please post!

[Paul_Yorks 97]

Not sure whether this helps but:my last machine was a quad core @ 3.3ghz, on a 8800 nvidia card. Ran pretty well until you added e.g. PNW to the equation. I ran unlimited, always, on frames, and had little stutter. But denser scenery meant I had to turn down AG or I was seeing flashing/texture corruption/stutters.New PC is an o/c i7 930 @ 4ghz and a 280nvidia. Naturally I cranked up the sliders. First thing I noticed - with FPS at unlimited - was MAJOR stutters. But nicer, crisper textures and more AG.So I conclude, and I admit this is pretty unscientific, that as you crank up the power, you tend to add to the scenery/autogen complexity, and you get nice images BUT the machine can't quite keep up with it all = stutters - until you cap FPS at e.g. 25/30, which lets everything work at a sensible rate so the stutters are fixed.My personal fix here is *******' BP=0 affair, which means almost stutter-free performance, at the expense of some blur, and using FPS = unlimited in the sim, with the external FPS limiter at say 25fps to slow down the stutters. Works pretty well, though in the end if you're flying a Spit at 300kts at 500', you'll see some issues. But this all seems only to work if you have the latest kit. With a pre-i7, all bets are off.Not sure if that helps!

[ryanbatc 12,930]

Not sure why - but I do get the same behavior your describe.However I have more stutters with unlimited so I use the FPS limiter bat file and much smoother.

[Kosta 122]

Not sure why - but I do get the same behavior your describe.However I have more stutters with unlimited so I use the FPS limiter bat file and much smoother.

Ryan, I believe you just misunderstood me just a little: my comparisons go with unlimited FPS in FS + FPS Limiter on the outside. I get much more stutters with external limiter than with the internal. My only problem with this scenario is, as everyone has it, lower FPS in every situation.It's like this:Internal FPS unlimited + FPS Limiter:- higher FPS- stuttering- blurries eventually, especially with photosceneryInternal FPS cap at 30:- lower FPS- no stuttering- no blurries, also with higher LODsI tried modifying all that comes to mind with bojote's formula for texture loader, where first of all, it doesn't have really much of an impact on any situation concerning how quick textures load. According to him it's just the upper limit, but not telling FS how quickly to load. Now, if we could discover how to make FS load textures quicker, that would be a bliss.

[Kosta 122]

Okay, now I removed JOBSCHEDULER and FFTF from my fsx.cfg, and suddenely, I'm getting higher frames. Apparently it's FFTF that makes a huge difference. JS is just barely couple of frames here and there (measured at over 70 frames).

[RelaxX 11]

What means FFTF?

[Kosta 122]

Haha this question pops at least once a week!Fiber frame time fraction tweak.

[Bert Pieke 7,585]

Internal FPS unlimited + FPS Limiter:- higher FPS- stuttering- blurries eventually, especially with photosceneryInternal FPS cap at 30:- lower FPS- no stuttering- no blurries, also with higher LODsI tried modifying all that comes to mind with bojote's formula for texture loader, where first of all, it doesn't have really much of an impact on any situation concerning how quick textures load. According to him it's just the upper limit, but not telling FS how quickly to load. Now, if we could discover how to make FS load textures quicker, that would be a bliss.

Try to compare CPU loads in the two scenarios, using Process Explorer or Task Manager showing each Core separately.I think you'll find that the Core that is doing the Texture scheduling (Core1 if Affinity Mask is set to 14) is loaded quite differently in the two cases.

[Kosta 122]

Bert, I totally agree with you! I definitely noticed that the load in the scenario with external limiter is way higher. I wonder why though? If, let's say, you yield to achieve 30fps, but situation in both cases allows for more than 30fps, and both limiters limit to 30fps, there should be same load... theoretically.

[Bert Pieke 7,585]

Bert, I totally agree with you! I definitely noticed that the load in the scenario with external limiter is way higher. I wonder why though? If, let's say, you yield to achieve 30fps, but situation in both cases allows for more than 30fps, and both limiters limit to 30fps, there should be same load... theoretically.

I have yet to hear an explanation for this.. BTW in my case the load for the internal limiter is way higher :( Based on trial and error, I find that if I can keep the settings so that I do not have a fully saturated CPUload, especially on Core 2 and 3 which do the texture loading, I get really smooth flight.KSEA with PNW is my worst case scenery, so that is where I tend to do my testing.Net result for me is LOD_radius at 3.5 and my system can keep up.FSX unlimited fps setting, FPS limiter set to 24.

[Kosta 122]

Errrrr, I gotta correct myself. I wrote the last post from the memory, which was apparently wrong. I went and tested couple of last minutes and noticed following:FSX Limited, internally: 100% on Core1Externally limited: around 40% on the Core1, and doesn't go any higherCore2/3 doing the same work in both casesNow, THAT is weird!Both situations tested just sitting on the airport, not flying. I'll do flying tests a bit later...

[Guest napamule]

HeeHee. The last 2 years I have read most of the suggestions out there. NickN's and Hay-soos's included. Tried them all. FPSLimiter, BP, FFTF, ETC. I leave some, and commented out some (but leave them there for later compare/re-install, with comments). I ALWAYS note what the poster's SPECs are as part of my observations. My conclusion is that we are fighting a losing war-BECAUSE WE ALL HAVE DIFFERENT SPECS!!!! Our MB are different. Our Chipsets are different. Our BIOS is different. Our Video Cards are different. Our Ram latency are different. And our sliders are different.What we need is to form a 'FSX RIG CLUB'. Where everyone in club has EXACT specs. Then go do some testing and do group tweaks (sort of like 'Lan' party, but for tweaking). This way (and maybe ONLY this way?) we would finally ALL get 30 FPS (consistently) with no stutters or blurries. Otherwise it's a futile endeavor and waste of time/effort IMHO.I am talking about drilling down to the resitor and capacitor type, copper content, etc. Also resistance and inductance values seen due to construction of MB. Down to the processor and type. Down to the 'heart' of any electronic circuit and the quality or method used to put it together. I bet all of our wires (gauge, etc) coming from our PSU are all different. Bet me. There in lies the vast differences we are faced with. NOT JUST THE SLIDERS AND THE CFG TWEAK LINES. When we say 'specs are all different' you are saying a mouthful as there is MORE to those differences than meet the eye.My specs? Pentium 4, dual core, 3.207 Ghz, BioStar MB, Bios=Pheonix v2.3 (2006), 2 Gig Kinston (1x2), Visiontek Radeon X1300 AGP 8X 512 Mb. I don't have FPSLimiter, get 30 FPS (locked at 50) and Orbx NW Blue runs at 25 FPS with AutoGen set to normal and I have some tweaks (BP=1000000, etc) but not all tweaks possible, and maybe some you don't have. My cfg has evolved on it's own, little by little, and is currently stable and solid as a rock. Want to see what tweaks I 'use'? (Your layout MAY be different (well, it WILL be different-please compare first). Might help you. Might not. So use/copy at your own risk (ha).FSX.cfg, as of April 28, 2010. cb ('Napamule').----------------------------------------------------------------[Display]TextureMaxLoad=30TEXTURE_BANDWIDTH_MULT=400UPPER_FRAMERATE_LIMIT=50WideViewAspect=True//ForceFullScreenVSync=TRUE //changed and moved to Graphics//ForceVSync=FALSE //changed and moved to Graphics[Main]DisablePreload=1FIBER_FRAME_TIME_FRACTION=0.10PerfBucket=7 //have 2 'PerfBuck', 1 here, and 1 in (Graphic)[DISPLAY.Device.VisionTek Radeon X1300 Series .0]Mode=1280x1024x32TriLinear=1[GRAPHICS]SHADER_CACHE_PRIMED=1693500672TEXTURE_MAX_LOAD=1024PerfBucket=7 //default was=6D3D10=0HIGHMEMFIX=1 //(nVidia card only?)STALE_BUFFER_THRESHOLD=2147483647ForceFullScreenVSync=1 //o=was TRUEForceVSync=0 //o=was FALSE[sCENERY]SmallPartRejectRadius=4[TERRAIN]LOD_RADIUS=4.500000TERRAIN_MAX_AUTOGEN_TREES_PER_CELL=1000TERRAIN_MAX_AUTOGEN_BUILDINGS_PER_CELL=1000//SWAP_WAIT_TIMEOUT=30 //not sure ??[bufferPools]Poolsize=1000000 //tried=5000000 //changed 10Apr07 ?RejectThreshold=98304CBNapamule

[Kosta 122]

Napamule, please stick to the topic. This is definitely an interesting observation, but this thread is about limited externally vs. internally. I'd like to keep it that way.As a sidenote, tweaking is called tweaking exactly for that reason, each one has to tune it's own engine to suit his needs, because as you say, no machine is same. But to understand what each tweak does, it has to be compared over vast amount of machines. Only then we'll find what we can use to benefit our OWN situation. And this is for the reason that there is probably not going to be ONE same machine out there. It would be like wining a lottery.So now, let's go back to this topic. It would help if you could check if FPS also go up if you unlimit it...