What always amazes me is how quickly people are able to determine (or think so) an increase or decrease in FPS by, let's say 2-3 FPS for instance.

A new addon comes out, a new tweak has been applied, a fixer installed, etc. and immediately the effect on FPS seems to be noticed by people.

 

I use 3 19 inch screens with matrox triple head and TrackIR

 

That means that the FPS hugely depends on where i look: at the pedestal, at the overhead, out of the window to look at at the scenery, or a mix of VC and scenery.

It depends on the zoom factor used as well.

 

All that means the FPS can differ from very low to very high. So, how on earth can you determine a gain or loss of 2-3 FPS ?

I'm going to preface my commentary with a fairly large disclaimer Avhpilot.  I sometimes seem to have a viewpoint more than a little contrary to that of the mainstream.  With that said and out of the way...

 

How are they so sure they are gaining (or losing) 2-3 FPS??  In a nutshell--they can't.  They are, to some extent, deluding themselves. 

 

Did you ever go to an auto parts store and look at all the cans of Wonder Product, each guaranteed to increase/improve "your fuel mileage" up to (some unbelievable) percent??   Usually somebody you know buys the product and finds an amazing increase in fuel mileage and so you buy into the scheme, add it to your tank, and find...nothing changes at all.  And truth be known, nothing changed for the other guy either.  He talked himself into the idea that things actually did change, or made some unconscious change in his driving style that resulted in an actual improvement, or sadly, simply talked himself into believing an improvement actually occurred.  Or just didn't want to admit being taken advantage of.

 

Do you remember a few years ago when it became fashionable to replace the air in your tires with helium??  Supposed to increase your fuel mileage to (some unbelievable) percent (sound familiar??).  Make your family safer from the possibility of tire fires (yeah, I never heard of them either) and--yadda X 3??  I did some lengthy research into adding helium to tires with my Mercedes Benz dealer and my son-in-law (nitro fuel funny car racer) as well as on line (the all knowing source) and found that helium can indeed do well in tires, provided you are driving at 200 mph on a race track or running the 1/4 mile at under six seconds and over 230 mph, or even piloting a real world jet (not a sim model).  For you and me??  Save your money and buy a new sim airplane, or perhaps the next big thing that will improve your FR's by 2 or 3.

 

But me??  I'd buy the sim airplane first.  Standing by to be bashed, flamed, and otherwise taken to task for daring to express the obvious

FPS not near as important as smoothness in FSX graphics. 25 FPS and smooth is way better than 45 FPS and stutters.

FPS not near as important as smoothness in FSX graphics. 25 FPS and smooth is way better than 45 FPS and stutters.

 

I agree.

 

But 25 FPS is already a high number in my case

 

So, i do not mind getting 8 - 15 FPS as long as it's smooth ...

 

So, how do we measure smootheness, is there an objective way to confirm that which we assume to experience ? do we have a unit of smoothness ?

I think "smoothness" is an imprecise term and is based on your visual experience only.  Does the sim need to visibly stop while things load up (or whatever goes on behind my monitor and inside that big case down there in the floor)??  Am I pleased by the quality of what is going on on my monitor??  I've never heard of a unit of smoothness.

 

My FPS is locked at 30 and ranges from 10 (12?) to 29.8.  Truthfully there are very occasional stutters so my view is not always "smooth".  But then again my sim seems fairly reliable and I don't spend four hours tweaking and tinkering for every one hour I fly.  It does what I want.  It may not do what you want.

That means that the FPS hugely depends on where i look: at the pedestal, at the overhead, out of the window to look at at the scenery, or a mix of VC and scenery.

 

It depends on the zoom factor used as well.

 

All that means the FPS can differ from very low to very high. So, how on earth can you determine a gain or loss of 2-3 FPS ?

 

Quite simple, really... For testing purposes I have a few saved flights. Since I always and only fly in the VC all flights are saved while in the cockpit. I also use TrackIR and before I save the testflight I look where I want to look (which depends on the situation) and then I pause TrackIR and then I pause FSX. After that I save the flight. So the next time I want to load the testflight, I first make sure TrackIR is paused and when the flight has loaded things are EXACTLY as they were the previous time with the exact same view.

 

When the flight has loaded the sim is paused: this is ideal for checking FSAA and AF.

 

For testing of fps and smoothness I make sure the flights is saved with the above conditions (everything paused) but also I make sure that the plane is on autopilot and settled in flight before I save. I have a rather 'heavy' flight above London like this which I use to check texture loading, fps and smoothness.

 

Texture loading is checked by making screenshots at specific times (so I end up with comparable screenshots everytime) and comparing them afterwards.

 

Fps is checked by looking at the ingame fps counter BUT (important) I have changed what I see in the game using the FSX.cfg:

 

[TextInfo.1]

AverageFrameRate=1,1

FrameRate=1,2

LockedFrameRate=1,3

 

By default you will only see the FrameRate, which is rather useless, but the AverageFrameRate shows you the lowest dip, the average and the peak and the lowest and peak are updated every 30 seconds or so (never counted the time though). This makes checking fps a breeze and I can clearly and easily see of fps is lower or not. Not that I will be able to say exactly how much, because there is some randomness everytime but I DO see if certain settings result in lower fps.

 

Smoothess can only be tested by looking and is very objective.

 

The above tests are done with TrackIR paused so the view is the same and comparable every single time. And btw whenever a test requires me to unpause the sim, I at least let it run for a minute so you can really see differences. Sometimes the sim needs a few seconds to get going.

 

To test shimmering, I first unpause TrackIR and look around with the sim still paused and after that I also unpause the sim to see how things look when moving. Obviously shimmering is also very objective.

 

I don't often test with TrackIR actually active because you won't be able to do comparable tests. Only for things like shimmering and smoothness I use TrackIR but even for smoothnes it is already doubtful if you should do that.

 

All in all I can test various things (all things that need testing) very easily and quite precise.

Quite simple, really... For testing purposes I have a few saved flights. Since I always and only fly in the VC all flights are saved while in the cockpit. I also use TrackIR and before I save the testflight I look where I want to look (which depends on the situation) and then I pause TrackIR and then I pause FSX. After that I save the flight. So the next time I want to load the testflight, I first make sure TrackIR is paused and when the flight has loaded things are EXACTLY as they were the previous time with the exact same view.

 

When the flight has loaded the sim is paused: this is ideal for checking FSAA and AF.

 

For testing of fps and smoothness I make sure the flights is saved with the above conditions (everything paused) but also I make sure that the plane is on autopilot and settled in flight before I save. I have a rather 'heavy' flight above London like this which I use to check texture loading, fps and smoothness.

 

Texture loading is checked by making screenshots at specific times (so I end up with comparable screenshots everytime) and comparing them afterwards.

 

Fps is checked by looking at the ingame fps counter BUT (important) I have changed what I see in the game using the FSX.cfg:

 

[TextInfo.1]

AverageFrameRate=1,1

FrameRate=1,2

LockedFrameRate=1,3

 

By default you will only see the FrameRate, which is rather useless, but the AverageFrameRate shows you the lowest dip, the average and the peak and the lowest and peak are updated every 30 seconds or so (never counted the time though). This makes checking fps a breeze and I can clearly and easily see of fps is lower or not. Not that I will be able to say exactly how much, because there is some randomness everytime but I DO see if certain settings result in lower fps.

 

Smoothess can only be tested by looking and is very objective.

 

The above tests are done with TrackIR paused so the view is the same and comparable every single time. And btw whenever a test requires me to unpause the sim, I at least let it run for a minute so you can really see differences. Sometimes the sim needs a few seconds to get going.

 

To test shimmering, I first unpause TrackIR and look around with the sim still paused and after that I also unpause the sim to see how things look when moving. Obviously shimmering is also very objective.

 

I don't often test with TrackIR actually active because you won't be able to do comparable tests. Only for things like shimmering and smoothness I use TrackIR but even for smoothnes it is already doubtful if you should do that.

 

All in all I can test various things (all things that need testing) very easily and quite precise.

 

Well, that is an impressive strategy and by the looks of it you have really thought that one out very precisely. i hope you will get to de some flying in between testing too:), interesting point about the average FPS etc, i didnot know about that so certainly worth looking into that myself.

 

I am also quite sure that you are not representative of the majority of folk that jump into threads so often, so quickly stating to have lost or gained the few frames.

 

I am just wondering how fair a comparison between a saved flight and reloading that after a fresh restart of FSX would be and if variables outside FSX could be different at different times and would account for a possible difference in FPS rather than the actual change applied (the tweak, the fix etc) being responsible. But i'm sure you've got that covered. 

(Edit)  snipped.  Sorry, my input isn't as useful, didn't spot just how specific this thread was to TrackIR. (Edit)

I think "smoothness" is an imprecise term and is based on your visual experience only.  Does the sim need to visibly stop while things load up (or whatever goes on behind my monitor and inside that big case down there in the floor)??  Am I pleased by the quality of what is going on on my monitor??  I've never heard of a unit of smoothness.

 

My FPS is locked at 30 and ranges from 10 (12?) to 29.8.  Truthfully there are very occasional stutters so my view is not always "smooth".  But then again my sim seems fairly reliable and I don't spend four hours tweaking and tinkering for every one hour I fly.  It does what I want.  It may not do what you want.

 

Well, reason alos is that i have applied the DX10 fix and i read somewhere that it should increase your performance when running in DX10. Not necessarily in increasing FPS but overall smoother compared to running in DX9 with similar FPS. And althoug i think i have noticed this improvement in smoothness i want to be sure of an objective way to measure it. Because, as you say, it may all be placebo after all.

 

So, how is smoothness defined: the human eye wil ignore anything above 25 PFS anyway so a lock at 30 will suffice.

Sometimes the sim can stutter while the FPS indicator shows high FPS (>40), so as Jeroen already mentioned, the SHIFT-Z method to measure FPS is useless and we should use the average FPS value in the .cfg file. So i cannot help thinking that when people say, "low FPS but it is smoother" they are experiencing actually high FPS when the ingame indicater shows low values.

 

I mean, smoothness is an experience resulting from high FPS so the human eye is comfortable with the moving images. Stuttering means a periodic freeze frame and that will bring the average FPS down, but if the FPS outside those stutters is still very high, the average FPS including those stutters could still reach acceptable levels.

 

Unless there's something like 'motion blur' i cannot think of anything else than FPS that accounts for the experience of "smoothness"

I almost didn't enter this conversation when you started it out by asking about FPS.   Disclaimer!!  The following should not be considered as a comment about any of the posters to this specific thread:  Posts about frame rates are usually one of the most circular and largely non-productive of time wasters available to the dedicated (or even casual) flight simmer.

 

I was gratified to see the comments take into account the many variables in any one flight and how to avoid those many variables so as to be able to get as clear a picture of FR's as possible.  Good thinking guys who are more concerned with exactly what they are seeing instead of simply saying "this is better because it's the newest hot trick".

 

I suppose "smoothness" will always be a somewhat elusive term.  Beauty really is in the eye of the beholder.

This topic got my attention because I was unaware that TrackIR affected frame rates - well, I'm not a perfect idiot (no one is perfect) - I understand why it could but I was unaware that there was anything noticeable.  So - yesterday I paid attention to my frame rates for a while (usually don't since I reliably get 32.2 to 32.9 locked at 33) - and low and behold I might have noticed a drop to ~ 31.5 if I moved my head quickly.  The usual (instrument) scanning head movements did nothing.  Looking out the side window seemed to make the most difference, causing a drop a couple of tenths below my usual.

 

And, like Dan George, I usually stay out of the FPS discussions except to post that I basically ignore my frame rates and occasionally add that I locked mine at 33 because otherwise rates bounced around between 25 and 80 - and caused stutters- or should I say 'lack of smoothness'.  The same philosophy as Henry and Antoine expressed. 

 

Somewhere I read from one of the Gurus of enhancing FSX performance (? *******) that the biggest performance killer was the user themselves and all of their add-on scenery and other goodies.  I use REX because it enhances the weather effects and textures - weather being an important part of flying, fancy buildings etc. being not so important.

This topic got my attention because I was unaware that TrackIR affected frame rates -

 

I wasn't trying to imply that TrackIR itself was bringing performance down. If it does, it's really insignificant.

 

TrackIR enables you to look in many different directions within a relatively short time. It's like having many screens attached to your computer but you can only look at one of those at a time.

And because looking outside your window to scenery can produce higher FPS then looking at your pedestal i was wondering what point=of-reference is being used when calculating FPS (in other words, which view/direction, which zoom factor and so on). Because, just like Dan George, i am very sceptical about FPS discussions as well.

I use a procedure similar to Jeroen's but use FRAPS benchmarking to check the framerate performance. It gathers extremely detailed data on the framesrates. You need to unlock any framerate limiting settings and run the exact same scenario multiple times and take averages, but if you really want to know how a particular setting or feature is impacting FPS, this is a way.

I am just wondering how fair a comparison between a saved flight and reloading that after a fresh restart of FSX would be and if variables outside FSX could be different at different times and would account for a possible difference in FPS rather than the actual change applied (the tweak, the fix etc) being responsible. But i'm sure you've got that covered. 

 

 

I think I have that covered, yes.  ^_^ Well, as far as possible. First of all: when I save a flight for testing purposes, I don't test during that flight itself. Secondly: I ALWAYS quit FSX and load the saved flight from Windows (bypassing the FSX menu) in between every test and comparison. So I never ever test different settings only by switching out of FSX to Windows to change something and back again. So every test is done with a 'freshly loaded' flight. When I want to test performance as precise as possible, I also restart the entire PC between every test (and make sure I don't load other stuff before I test). For things like FSAA etc. this isn't necessary. 

 

All in all this DOES take up some precious time but it can actually be quite fun to do, specially when you discover surprising things.

 

However, even though I think I have a nice testing method, I discovered that TESTING has its limits...:

 

 

And althoug i think i have noticed this improvement in smoothness i want to be sure of an objective way to measure it. Because, as you say, it may all be placebo after all.

 

The best way by FAR to test smoothness is to simply FLY and USE the sim. Seriously. Testing has its limits!!! 

 

You tend to look closely at things that you normally don't look that close at when flying. It's very hard NOT to do this. This means that if you aren't careful you are 'testing' things that don't really matter for the overall experience when you simply fly!

 

A few weeks ago I tested DX10 and DX9 and came to the conclusion that DX9 performed just as well, if not better. I specially looked closely at stutters and smoothness using the methods described above and ended up choosing DX9. Shortly after that I bought a plane that had great VC shadows so I decided to fly it with DX10. Things were nice and all but after a week (!) I went back to DX9 for various reasons. Within no time I noticed slight stutters... which weren't there with DX10. I only really discovered this after actually using the sim for a week without thinking about testing!

 

No matter how well your tests are, some things can only be tested by not testing but just using the sim.   As I posted my tests are done with a plane on autopilot, quite often looking steadily in a fexed direction. Well, in normal use I never fly like that: I look around a LOT with TrackIR so the only way to really see how things perform is to just have fun with FSX (which is the purpose of the game anyway).