Framerates

[Aerosoft Team [Inactive Account] 51558]

After the discovery that keeping an unlimited framerate (so no locked framerate) makes your framerate up to 25% higher in almost any dense location, we now are investigating something new. We hope you could assist us with that and try it out.

Go to a location when you got about 20 fps (not so important actually) remember the average fps.

Now click World | Scenery Library and directly [OK] without doing anything on this screen. On some machines we see a much higher (up to double) framerate, on others we see hardly difference and one some we see a drop in fps. If you got a moment try it out and let us know.

[darem 106]

After the discovery that keeping an unlimited framerate (so no locked framerate) makes your framerate up to 25% higher in almost any dense location, we now are investigating something new. We hope you could assist us with that and try it out.

Go to a location when you got about 20 fps (not so important actually) remember the average fps.

Now click World | Scenery Library and directly [OK] without doing anything on this screen. On some machines we see a much higher (up to double) framerate, on others we see hardly difference and one some we see a drop in fps. If you got a moment try it out and let us know.

May I add something? This locking framerates vs. unlimited frames seems to be related to the type of card you have.

With my Geforce 8800GTX Ultra (768MB), unlimited framerates introduced major stuttering. Locking the framerate at a reasonable point smoothed everything out.

With my new ATI 4850 (1GB), unlimited framerates provide smooth gameplay and excellent rates. Locking the framerate makes the game unplayable. No framerate lock: 35fps on average, with lows in the mid 20ies. Framerate locked (no matter where): 15fps. Or less.

BTW, Mathijs, please revise your recommendation for a graphics card as soon as possible. I bought the ATI because of your recommendation, and it was a waste of money. The 4800s (with the exception of the top of the line product) have huge difficulties to render clouds - if you choose simple clouds, the card flies, if you choose the better clouds, it creeps. Where you have 30fps with simple clouds, you will have 8fps with good clouds and an overcast sky. This makes products like FEX completely unuseable. The cloud rendering problem is noticeable in FS9, too, but much less of a hassle. In X-Plane 9, clouds have little effect on the performance of the 4850.

It's not a thing that is on my computer only, it is widely discussed on various boards. It is a problem that has been present with every catalyst driver, even the latest one. Sorry to say, but unless you buy a 4870 X2, NVidia ist the way to go for FSX.

Just to let you know...

Bernhard

[Rafal Haczek 1612]

keeping an unlimited framerate (so no locked framerate) makes your framerate up to 25% higher in almost any dense location

For some people/systems/configs it may be true, but reasonably locked frames (25-30) give the system spare resources to load textures better which helps with ground blurries and (32bit especially) textures load times for example. At least that is what I read and what I think I experience with my MFS.

Best regards,

Rafal

[Aerosoft Team [Inactive Account] 51558]

For some people/systems/configs it may be true, but reasonably locked frames (25-30) give the system spare resources to load textures better which helps with ground blurries and (32bit especially) textures load times for example. At least that is what I read and what I think I experience with my MFS.

Best regards,

Rafal

Yes that is what I always thought and what I always advised. But unless you are very high up it just doesn't work like that. We now tried it on nearly any machine we could find and we see the same on every machine. As soon as you allow FSX to run unlimited it will reach a higher FPS without loosing any of the fluidity. In simple terms with FPS capped at 24 you might not reach it, but with unlimited you might reach 28. We honestly believe there is a rather serious bug in the fps limiter.

Now the whole FPS issue between FS2004 and FSX is complex because most people think FPS actually means a lot. But FSX runs the frames rather fluid where FS2004 always had 6 frames fast and 3 frames slow and that caused FS2004 to get good fps numbers even though FSX might look a lot smoother with much slower fps. And we were afraid the unlimited fps would change this. But even at unlimited we see only a 1.7% variation between fps so it remains smooth.

Let's put it differently. FSX is not a hobby for me but work. I run it 25 hours a week. I'll never use the fps limiter again.

[Aerosoft Team [Inactive Account] 51558]

May I add something? This locking framerates vs. unlimited frames seems to be related to the type of card you have.

With my Geforce 8800GTX Ultra (768MB), unlimited framerates introduced major stuttering. Locking the framerate at a reasonable point smoothed everything out.

With my new ATI 4850 (1GB), unlimited framerates provide smooth gameplay and excellent rates. Locking the framerate makes the game unplayable. No framerate lock: 35fps on average, with lows in the mid 20ies. Framerate locked (no matter where): 15fps. Or less.

BTW, Mathijs, please revise your recommendation for a graphics card as soon as possible. I bought the ATI because of your recommendation, and it was a waste of money. The 4800s (with the exception of the top of the line product) have huge difficulties to render clouds - if you choose simple clouds, the card flies, if you choose the better clouds, it creeps. Where you have 30fps with simple clouds, you will have 8fps with good clouds and an overcast sky. This makes products like FEX completely unuseable. The cloud rendering problem is noticeable in FS9, too, but much less of a hassle. In X-Plane 9, clouds have little effect on the performance of the 4850.

It's not a thing that is on my computer only, it is widely discussed on various boards. It is a problem that has been present with every catalyst driver, even the latest one. Sorry to say, but unless you buy a 4870 X2, NVidia ist the way to go for FSX.

Just to let you know...

Bernhard

Mmmmm... We have not seen that problem with the 4800's but I agree there have been more reports. But as prices changed i was about to prefer another line of cards anyway. What would you (obviously you have an opinion and I like that!) advise at this moment for a machine that should stay under 1000 euro in total (ex screens).

I got the same Geforce 8800GTX Ultra (768MB) as you got an even unlimited fps I get smooth fps, capping the fps only helps when I am at cruise altitude, makes it smoother overall. The moment I get close to airports unlimited fps just gives me 20% more fps. Shows how varied these things can be. If you are not running a quad core I would think that the CPU starts to cause problems in your machine.

[darem 106]

Mmmmm... We have not seen that problem with the 4800's but I agree there have been more reports. But as prices changed i was about to prefer another line of cards anyway. What would you (obviously you have an opinion and I like that!) advise at this moment for a machine that should stay under 1000 euro in total (ex screens).

I got the same Geforce 8800GTX Ultra (768MB) as you got an even unlimited fps I get smooth fps, capping the fps only helps when I am at cruise altitude, makes it smoother overall. The moment I get close to airports unlimited fps just gives me 20% more fps. Shows how varied these things can be. If you are not running a quad core I would think that the CPU starts to cause problems in your machine.

Funny, the stuttering was prominent on the Dualcore AND the Qaudcore for me. I have accumulated a history of heavy upgrades during the recent months *g*, so I know a lot of system configurations by now.

At the moment, I am running a machine close or even equal to your specs: Q9550 (now overclocked to 3,2GHz), 4GB Ram, Vista64, Ati 4850 1GB, 10.000rpm drive for system, separate drives for FSX and again for supporting programs. The machine is extremely fast as long as I don't turn on detailed clouds, with detailed clouds, it's unplayable with ovrcast skies or once you fly into a cloud.

Which GC would I recommend in the price range specified? Generally speaking, NVidias seem to handle FSX better, especially when it comes to weather. 512MB VRAM are a must, more is better UNLESS you have a 32 bit OS - VRAM then will be deducted from the available memory, so this is a tradeoff. But from my point of iew, you can't go wrong with Vista64 by now. Stable and fast. So with a 64bit OS, I would recommend a 1GB GC. VRAM speed is another thing to consider, especially if you have a fast processor.

In my opinion one has to balance the system in the best possible way. Our 8800GTX Ultra is bored with a 2.6GHz Dualcore, and the fastest processor will not help if you use a 7600 with high resolution. 9800s are basically relabeled 8800s, so if I would have to recommend a card, it would be a 8800 with fast VRAM and 1GB. If you don't need detailed weather, the 4850 is excellent, and if you have more money than you can possibly spend, I would go for the best 280 you can get *g*. The 4870 X2 is reported to be fast enough to not show the weather problems, and MAYBE these issues can be solved in a newer Catalyst driver, but I wouldn't bet on that - FSX is not a milestone for card manufacturers to be mentioned for...

[thomas29 5]

Hello Mathijs,

I have also got the bug with the Framelimit, it works well with 20-25 FPS in not dense areas,but in dense areas like Flytampa Kai Tak, EDDF, Heathrow..it is better than not to limit the frames with my System, and the frames are fluent. The other problem is, if I set the framerate unlimited and fly around for example through the VFR Germany West scenery (not dense areas), I have got mayor stuttering, FPS vary then from 25-50. So I think its not an ATI Problem. Best would be the possibility to turn on ( big airports...) and of the Framesettings with a Key combination. Anyone an idea ?

Many Greetings Thomas

[darem 106]

Ok, tried what you asked for:

Before: 34fps on average

After: 34fps on average

Tried in a Cessna 172 over Vienna in Austria Professional X with autogen set to very dense. System as described above.

[limp 29]

I just tested what you asked... And well... I get FPS over 20 in Heathrow with the FA18...

I didn't try flying yet, but the result seems very promising... I had the false idea that locking it at 24 fps was a good thing...

I'll give it a try when I have more time. Thanks for the heads up

[Snave 466]

Tested as requested over LFBL, Bellegarde in France in the default 172. default scenery, UTX off, only replacement textures Natural Trees V 2.0

27 fps before

27 fps after

For researches sake I limited the fps to 25 and tried the test again, to see if there was any implied or apparent increase/decrease in fluidity in frame display.

There wasn't.

However one interesting anomaly I observed was increasing the texture detail from 2m to 1m actually resulted in an increase in fps when the fps limiter was set to unlimited. I don't recall seeing this before, but it might also indicate the fps limiter is having an effect outside of its intended role.

What is important to remember though, is that those `missing` frames when the fps limit is activated are an indication of CPU and/or GPU time being used elsewhere other than the immediate display of more-frames-consecutively-in-the-same-time-period.

Although fps is a handy indicator, it does not indicate the workload of the sim, and it is entirely likely that the fps limiter has an effect of transferring pure fps cycles to other areas of the sim for greater fidelity elsewhere. I have noticed, but not measured, that the speed of rotation and response at take-off rotation does `feel` different when various fps limits are applied. the whole rotation and wheels-off is smoother not with ever higher frame rates as one might suppose, but between a fairly narrow band on the fps limiter - 20-25 in my case. Above or below that figure the fluidity, the sense of motion in a fluid and dynamic environment, is reduced. Anything above 30 on my rig, and it feels like there is control lag, which is a contra-indicator to what one might expect.

This may simply be an individual bottleneck on my ageing computer, but it might be some kind of algorithm is being applied to the Fiber, defining in a behind-the-scenes way what the FSX .cfg setting

FIBER_FRAME_TIME_FRACTION=

does specifically, and in a user-configurable manner for scenery texture display, but for other aspects of the sim running through the Fibers.

If that is the case, the reduced fps may be preferable, in which case the fps limiter may be doing exactly what it was intended to do, allowing passthrough on the pure generating of frames in favour of CPU cycles/Fiber time for non-display essentials like atmospherics, flight dynamics and AI.

[Snave 466]

To clarify my previous post, here is the definition of Fibers and the FPS Limiter, as supplied by Adam Szofran

...Further, setting it to values greater than the default value on single-core machines can increase frame rate volatility because it increases the amount of time we *might* allocate to fibers if there is adequate work waiting in the queue. When the queue is full, we’ll allocate the full amount of time to fibers but when the queue is empty, fibers get no time because there is no work to do. If the rate of new jobs entering the fiber queue is bursty and the full time allowed for fibers is large, then you can imagine how this would increase volatility.

If people feel like the fibers aren’t getting adequate resources, they would be better off leaving the fiber fraction alone and just lowering the frame rate limit slider, which helps divert more CPU time to primary thread fibers without increasing volatility.

Although this is an answer to a different question about setting large numbers well above the default Fiber_Frame_time_Fraction of 0.33 I think it also provides an insight into why the fps limiter is doing what it is doing. First of all there is the implied statement that although the sim runs on Fibers, there is/are a Primary Fiber/s which must, by implication take CPU priority. It may also take GPU priority, but that's a different question. The fps limiter does divert more CPU time to primary thread fibers, that's a stated fact and I see nothing to disagree with that in the operation of the sim, but it's getting that CPU time how? It MUST be robbing Peter to pay Paul, and therefore the conclusion is that while you may run higher fps with the UNlimited setting, you are achieving this at a cost to the sim elsewhere.

The smoothness apparent at higher fps may simply be the by-product of the sim skipping, or short-changing in some other process that would otherwise have been given it's `slice` of Fiber time priority. It might be flight dynamics, it might be weather dynamics, it might be AI, but it is somewhere. The FPS limiter may well be doing its job on unseen factors that we cannot easily measure.

The effect of the fps limiter should be gauged in the light of CPU and RAM usage if you wish to test whether it is having an undesirable effect, not smoothness or fps in the sim.

[limp 29]

Works for me like a charm

I'm going to get Tahiti X and the F16 very soon

[mopperle 4162]

Hi,

very nice screenshoots, but they don't answer Mathijs question: average FPS before and after

Otto

[SF22 5]

Hi! I tried it. The FPS increased from about 18-20 FPS to about 30 FPS in spot view but if you change the view they will return to the normal even if you return to the spot view.

My system specs: AMD Athlon 3800, a NVIDIA 7600gs 512mb video card and 3gb of ram. (I know, I have to buy a new computer)

[mopperle 4162]

After the discovery that keeping an unlimited framerate (so no locked framerate) makes your framerate up to 25% higher in almost any dense location, we now are investigating something new. We hope you could assist us with that and try it out.

Go to a location when you got about 20 fps (not so important actually) remember the average fps.

Now click World | Scenery Library and directly [OK] without doing anything on this screen. On some machines we see a much higher (up to double) framerate, on others we see hardly difference and one some we see a drop in fps. If you got a moment try it out and let us know.

Did the test and no change in fps.

Otto

[Cody Bergland 4]

I'm going to test this when I get home.

Regarding capped FPS, I have always, and will always continue to say that unlimited is the best option for most folks out there. After spending numerous hours working with FSX and filming, I have found this statement to not only be true, but an absolute must for my work. I attain an easy 5-10 FPS gain by doing this with no perceivable loss in smoothness, the result of more blurries, etc. I highly recommend this.

I'm running a Core 2 Duo e6400 overclocked to 3.1ghz with a Nvidia 8800 GTX 768.

[Snave 466]

I'm going to test this when I get home.

Regarding capped FPS, I have always, and will always continue to say that unlimited is the best option for most folks out there. After spending numerous hours working with FSX and filming, I have found this statement to not only be true, but an absolute must for my work. I attain an easy 5-10 FPS gain by doing this with no perceivable loss in smoothness, the result of more blurries, etc. I highly recommend this.

I'm running a Core 2 Duo e6400 overclocked to 3.1ghz with a Nvidia 8800 GTX 768.

Saying so implies you know what is missing when you go for `frames at all costs`. So do tell.

[Andy B. 40]

I'd just like to add that if you have a third party traffic package installed when you go to the scenery library and don't change anything but just click on OK this reloads the traffic BUT it doesn't delete any traffic that already existed. So in essence you can end up with 2 of every flight. If you go to the scenery library again and repeat the process you will end up with 3 of everything and so on and so on. If you cancel out of the scenery library manager instead, then there is no change. Obviously the more flights you have up in the air the more stress its going to put on your CPU.

I've used unlimited frame rates for pretty much the whole time since the release of SP1 and SP2 for FSX, as my frame rates have always been better this way, and this is with 2 very different machines, one being an AMD rig and the newer rig being an Intel machine.

With the newer Intel rig in XP, if my frame rates averaged over 45 FPS then I would see stutters believe it or not, anything below this down to about 22fps was smooth as silk. So in area's where I averaged 45+ I would lock the frame rate down to 30 or so and the stutters would disappear again. As soon as I moved to an area where 45 FPS aren't viable with an unlocked framerate my machine would start to struggle and FPS would dip dramatically to around 15-18 FPS, the instant I unlock the frame rate though it would climb back up to 30+.

Now to muddy the waters a little more just recently I've given Vista 64 another go and this time made a completely separate and totally clean install of FSX. In the past I'd tried using the same install of FSX in XP and Vista and it worked fine but was significantly slower (up to 50%) in Vista than in was in XP. Well since the reinstall things have actually swung the other way for me, frame rates are probably about 15-20% better in Vista and more significantly when frame rates dip into the 18-20 area things are very much smoother, it actually feels more like 25+ FPS, and I don't get stutters when the frame rate climbs to 45 or more either, which is great.

The other major advantage for me now though with Vista 64 and 4 gig of RAM is that I never see any OOM errors anymore. In XP this was starting to be a significant problem with some of the very detailed add-ons like the PMDG MD11 and VFR London etc when you turn everything up to the max.

FSX is a funny beast indeed and I too believe that there is a major flaw in the frame rate limiting code somewhere with certain hardware combinations.

[Cody Bergland 4]

There is nothing to tell. Check my videos. They tell the story for me

When you spend 40+ hours a week filming in FS, you notice such things as you are constaly staring at an FPS counter in FRAPS.

[Snave 466]

There is nothing to tell. Check my videos. They tell the story for me

When you spend 40+ hours a week filming in FS, you notice such things as you are constaly staring at an FPS counter in FRAPS.

And clearly miss the myriad other things going on - or not going on - invisibly. Fortunately there are those who can see beyond the blatantly obvious and can look at the simulation product as more than merely pretty animated pictures. A painter can paint a picture of an ill man. It takes a surgeon to get under the skin and remove the appendix.

There is no doubting that spending too long staring at the FPS counter is bad for you. MS ACES tried to make this point from the outset and indeed, if you make used of the presets ACES provide they offer sustained fps limits below what even we suggest. That tells me that something is being lost, not gained, to give you your prettified pictures and that their threaded philosophy - fibers - actually require a lower-than-we-might-wish (or expect) frame rate to do both the primary and other fibers justice. Time is the fixed constant. CPU and GPU power is finite. Increase processing of rates of frames and you must lose something, somewhere else. After that it is purely a matter of personal taste whether those limitations intrude on your sim experience or not. Those unconcerned with system fidelity, flight model appreciation and complex computation of variables may well enjoy the enhanced-fps experience. But there is no gain without pain.

A simple point to note: The obvious and annoying jump in barometric pressure between adjacent weather areas when using FS Real Weather is much mollified when the fps is lower. When you spend 40+ hours a week actually analysing and studying the various sims, as well as playing them, you notice such things when you are staring at a staccato gauge response in the VC.

[darem 106]

An addendum to the ATI/clouds issue: I just found out that you CAN use detailed clouds as long as you don't use AA.

[Cody Bergland 4]

Snave, I don't understand what you are getting at. I suggest that doing so provides no measurable loss of simulation experience. My video work proves my point... yet somehow you say that what I am getting is a fairy tell yet have no proof to back it yourself?

My work goes far beyond making videos and pretty pictures.... However, defending my professional background or work is not the scope of this conversation. If you can somehow point to how I am somehow missing something by setting my FPS to unlimited, by all means prove me wrong. Since the release of FSX I have been utilizing this tweak, and in that time I have not noticed nor had any evidence provided that has suggested otherwise.

And the FRAPS FPS counter is much more accurate than the built in FSX one.

And to use your own words:

The FPS limiter may well be doing its job on unseen factors that we cannot easily measure.

If the job it is doing is unseen (as in not visually seen, felt and it doesn't effect anything measurable that we know about), then what is the harm? If it smells, looks and flops like a fish, it is a fish. Until someone provides evidence of otherwise that is confirmed by another party, I will continue preaching that for most people, this is a good and simple tweak.

[Waleed 29]

I am no expert, and sadly lately am too busy to try too many things but:

I followed the advise given by Mathjis and Shaun for the oom fix, and I succesfully flew the payware hunter with no oom, but with some stutter over london.

I have london vfr, mega heathrow, but no ai running.

I moved my slider to unlimited, and flew an rj170 from eglc to egll with a trip around london vfr area with no stutters at all.

I had tcass, and weather and turbulance on, but no flight plan in the system with updated weather(fsx).

I am no expert and do not even know how to look for the frame rates, but in my case, the unlimited frame rates seems to have helped in giving me a smoother flight.

[Rafal Haczek 1612]

I am no expert and do not even know how to look for the frame rates

Shift+Z (repeated)

Best regards,

Rafal

[Snave 466]

If the job it is doing is unseen (as in not visually seen, felt and it doesn't effect anything measurable that we know about), then what is the harm? If it smells, looks and flops like a fish, it is a fish. Until someone provides evidence of otherwise that is confirmed by another party, I will continue preaching that for most people, this is a good and simple tweak.

Obviously you base your assessments on visuals. They are not the only criterion in use in a flight simulation product. There is little point in continuing this discussion with you. You tell me you want to make your assessment of success entirely on visual perspicacity. That's your choice. Enjoy it. I look deeper. I enjoy that. If it smells, looks and flops like a fish, it might be a dolphin. Which is a mammal. Look closer and find out, or just call it a fish and have done with it. That is also the basis of what FS is all about.

The basis of this forum thread is whether the fps limiter is having an undesirable effect in FSX, by artificially bringing fps levels to below what they `should` be. The evidence so far suggests that when the fps limiter is set to unlimited, thereby providing supposedly a better visual experience (although the evidence seems to neither confirm nor deny this so far) doing so is at a cost. Whether that cost s outweighed by any benefit is a matter for research and interpretation.

There is no such thing as a free lunch.

A 20% uplift in frame rates MUST be equated to a similar uplift in CPU processing, RAM access or GPU power demand, or even a reduction, if the fps limiter was somehow a bottleneck and removing it frees up the sim to be all that it can be. The fact that it isn't happening - no significant change at all in CPU, RAM or GPU demand - suggests that FSX is cutting back on one thing, in order to provide prettified pictures. The Szofran statement suggests that you can interrupt a Fiber. The FPS limiter interrupts the, for want of a better word, `prettified picture primary fiber`, and so allows other fibers to compete for CPU time. What those fibers DO may have little or nothing to do with visuals in the direct sense, but they will, individually and cumulatively, engage on some level with the FSX experience.