Jump to content

  •  


   
Where Flight Simulation Enthusiasts Gather from Around the World!

HOT SPOTS:
 rarrow.gif The 2014 Demographic Survey! - NEW!

rarrow.gif What's new in the Library

rarrow.gif Member Blogs

rarrow.gif Member Reviews - Contribute a Review!
rarrow.gif Bargain Hunters Forum
rarrow.gif Classified, Want, Swap Ads

rarrow.gif AVSIM Team Speak System

rarrow.gif Video Library
rarrow.gif Image Gallery
rarrow.gif Community Links Library
rarrow.gif Tutorials

rarrow.gif FS9. FSX & P3D CTD Guide

rarrow.gif FSX / P3D Configuration Guide

rarrow.gif AVSIM Publications 



Simulation's Premier Resource!

AVSIM is a free service to the flight and simulation communities. Please help us keep it that way. Donate what you can today! Thank you for your support!











. . .
Photo
- - - - -

FFTF=0, tweak against FPS drop?


  • Please log in to reply
199 replies to this topic

#1 Kosta

Kosta

    Member - 8,000+

  • Members
  • 9,086 posts
  • Joined 11 Years and 17 Days Ago.

Posted 28 April 2010 - 04:30 PM

OK guys, some testing is required please, so anyone who might feel inclined to help a bit and maybe see how this works out:All normal tweaks incl. BP=0, are allowed. VSync tweaks should be omitted for this test!! (otherwise those will cap to 60fps)For the first test leave out FIBER_FRAME_TIME_FRACTION out in the fsx.cfg (or //// in front of it). FIBER_FRAME_TIME_FRACTION = FFTF.OK so let's begin:1) Save a flight with a default aircraft at an airport of your choice (somewhere where you usually get higher fps, so that measurements are more accurate) - important: without clouds and in static view, outside or inside. Here it's simply important that the situation is something you can reload and it will ALWAYS give you same FPS with same settings! Important though, it should NOT be a situation in which you get more than 90fps, ideally. I have a test situation which gives me 73fps.2) Run FSX normally, without FPS Limiter, set unlimited FPS and read out the FPS.If everything is working normally, and you reload the flight 2-3 times, you will always get this same FPS readout. Remember it.3) Now, limit the FPS to 99, and if everything is right, FPS should drop by a fair amount - we are talking some 20-30%. Mine in that case drops to about 53fps.4) Next, enable the FFTF tweak in the fsx.cfg and set it to:[Main]FIBER_FRAME_TIME_FRACTION=05) Load the test situation and check the FPS. Mine went back to 73fps.I would be very grateful if some could do this small test, maybe we found another useful setting for our beautiful hobby :(

Help AVSIM continue to serve you!
Please donate today!

#2 Illegitimi non carborundum

Illegitimi non carborundum

    AVSIM Staff Reviewer

  • Reviewer
  • 1,025 posts
  • Joined 9 Years, 3 Months and 16 Days Ago.

Posted 28 April 2010 - 08:31 PM

KostaThis is quite an important test, and I guess that nearly all of the testers will have an LCD monitor with a basic refresh rate of 60Hz.Now it would be interesting to carry out your test on a LCD monitor capable of a RR of 120Hz or if anyone has got one a RR of 240Hz. (In theory you wouldn't need VSync enabled with these types of monitors, unless the video card supports a huge Refresh Rate).There are also LCD TVs out there with RR of 200 and 300Hz and it would be interesting to see if they get different results.I also wonder if anyone has got a superfast CRT with a RR of >60Hz and what sort of results that they can get.It's a great pity that we don't know without a lot of digging the capability of our video cards in what is the maximum RR that they can support.Great Test - looking forward to the results.PeterH

#3 Kosta

Kosta

    Member - 8,000+

  • Topic Starter
  • Members
  • 9,086 posts
  • Joined 11 Years and 17 Days Ago.

Posted 29 April 2010 - 12:20 AM

I just had a thought as I read through the older thread, maybe it is also relevant. Someone reported getting blurries with FFTF=0.It is crucial you have a quad core machine for this test. And also probably important to have following settings:[JOBSCHEDULER] <- VERY IMPORTANT, otherwise will turn the balance off and give you extreme blurries if you don't offload the main scheduler to the Core1AffinityMask=14[Bufferpools]Usepools=0[Display]UPPER_FRAMERATE_LIMIT=30TEXTURE_BANDWIDTH_MULT=40TextureMaxLoad=15[Main]FIBER_FRAME_TIME_FRACTION=0DisablePreload=1[GRAPHICS]ForceFullScreenVSync=1ForceWindowedVSync=1HIGHMEMFIX=1ALLOW_SHADER_30=1SHADER_CACHE_VERSION=11STALE_BUFFER_THRESHOLD=1024TEXTURE_MAX_LOAD=1024[SCENERY]MAX_ASYNC_BATCHING_JOBS=2[TERRAIN]TERRAIN_MAX_AUTOGEN_TREES_PER_CELL=4500TERRAIN_MAX_AUTOGEN_BUILDINGS_PER_CELL=3000LOD_RADIUS=4.500000AA and AF in the FSX should be set to ON. Nhancer should be set to Enhance and with settings as you wish (I use 8xS + 16AF), Preload is set to 1 (not at default 3).After it's working with these settings, you can most definitely set LOD_RADIUS=7.5. Which of course requires a bit of a muscle, but YMMV.I said before settings are as you wish, BUT, they probably aren't. It works here as it works probably because of those settings.

#4 bojote

bojote

    Member - 1,000+

  • Members
  • 1,565 posts
  • Joined 12 Years, 4 Months and 28 Days Ago.

Posted 29 April 2010 - 12:29 AM

Tested it with my laptop, you are most definirely right, however.. in the laptop the blurries are very evident, I need to try on the desktop, but will do that tomorrow.
******* (Hey-sus) &nbsp;&nbsp; Posted Image

Automated Tweaking & Tuning tool for FSX here &nbsp;|&nbsp; Download the Shader 3 Mod for FSX (w/ autoinstaller) here &nbsp;|&nbsp; My other add-ons can be found here

Intel i7 920 @ 4.2Ghz - 6GB 7-7-7-16 @ 1603, GIGABYTE GTX 480 @ 800/1600/1900 Win7 Ultimate 64-Bit w/FSX SP2, REX2, GEX, UTX-USA, ORBX PNW, FSGenesis Mesh + USA Landclass, ASE, FSPassengers, Level-D 767 and SquawkBox + Ultimate Traffic 2 (SP1) SMS File (100% VATSIM flyer in VC w/TrackIR4)

#5 BAW303

BAW303

    Member

  • Members
  • 96 posts
  • Joined 5 Years, 8 Months and 24 Days Ago.

Posted 29 April 2010 - 02:57 AM

KostaI have Tested your theory as you set out above, and Im happy to report you are correct about this.I have taken this one step further, and tested with my saved default test set up: JS41, FSDT JFK, Cloud cover, and fully tweaked .cfg. The result was exactly the same.adding FIBER_FRAME_TIME_FRACTION=0 frames are locked at 30fps, and we are now able to achieve 30fps external and 25-30fps in the VC.......Well done my friend...!!EDIT: hmmmm... there maybe some texture issues present with this tweak in, some of the ground scenery & textures are not being rendered correctly, will do some more testing
Craig Sells

#6 J van E

J van E

    Member - 5,000+

  • Members
  • 7,229 posts
  • Joined 8 Years and 16 Days Ago.

Posted 29 April 2010 - 04:20 AM

Haven't tried it (yet) but I know that FFTF lower than the default 0.33 gives blurries on my system: with FFTF at zero no time is allocated at all to load texture. It seems to me that this test may be nice for getting good fps while standing on an airfield, but it looks rather useless for actual flying...? What exactly do you hope to achieve with this test? Is it just a test to check out if fps goes back to (in your case) 73 or should it help us with getting FSX to run even better?

With regards,

Jeroen


#7 BAW303

BAW303

    Member

  • Members
  • 96 posts
  • Joined 5 Years, 8 Months and 24 Days Ago.

Posted 29 April 2010 - 06:56 AM

After further testing my reports of texture problems were misplaced, I cannot reproduce the issue again.I am also not seeing any more or less texture bluring, with or without this tweak added.What I am seeing now, is an incredibly smooth sim locked at 30fps with NO drop off in performance.... :(
Craig Sells

#8 Inactive Member_rdg_***

Inactive Member_rdg_***
  • Guests
  • Joined 44 Years, 7 Months and 26 Days Ago.

Posted 29 April 2010 - 10:03 AM

Hello,There had been considerable discussion on FFTF with ACES (Phil Taylor) and of course many in the community. The link below may be something you wish to read. At the very least, it provides a little more information to help you decide. I personally do not use it but again like so many tweaks, "your mileage may vary"Also please note that if you have a really speedy well tuned machine, you may not notice any change from min to max. What many tweaks have in common is that if a value is set that is lower than or greater than the allowed value, FSX will set this to a default level after the variable is read into the FSX executable. I have no information as to what happens with this value internally when it is set to zero. I is my understanding that this tweak has the greatest affect on single core machines.http://blogs.msdn.co...f-the-week.aspxRegardsBob G

#9 BAW303

BAW303

    Member

  • Members
  • 96 posts
  • Joined 5 Years, 8 Months and 24 Days Ago.

Posted 29 April 2010 - 10:50 AM

Bob,It would indeed be nice to know what goes on in FSX with fftf=0, because what this does achieve, is removing the 20-30% decrease in performance you receive when using the internal limiter. It is not without its drawbacks though.. a few blurries have crept in, but not excessively so.......so far. I can imagine this maybe a lot more of an issue with orbx airport sceneries.
Craig Sells

#10 Inactive Member_rdg_***

Inactive Member_rdg_***
  • Guests
  • Joined 44 Years, 7 Months and 26 Days Ago.

Posted 29 April 2010 - 11:20 AM

Bob,It would indeed be nice to know what goes on in FSX with fftf=0, because what this does achieve, is removing the 20-30% decrease in performance you receive when using the internal limiter. It is not without its drawbacks though.. a few blurries have crept in, but not excessively so.......so far. I can imagine this maybe a lot more of an issue with orbx airport sceneries.

Every application or most read values from a cfg file and yes the value of zero is not known, at least I dont know. Some have said .01 to .99 are valid values. Whether the code in the FSX excutable sets this to a default, I do not know. As far as ORBX, one thing you do know is that the amount of texture data being read is tremendous with this addon depending on the slider settings in FSX. Logically the more texture data being processed the greater the chance of the blurries. On the other hand, the better the hardware the chance of blurries decreases, if of course the tweaks used are used properly and testing is done systematically. You cannot count the number of times the word "placebo" and "your mileage may be different" has been said on this and many other FS forums, in respect to tweaking. I just upgraded my hardware a couple of years ago and am getting ready to do it again. RegardsBob G

#11 bojote

bojote

    Member - 1,000+

  • Members
  • 1,565 posts
  • Joined 12 Years, 4 Months and 28 Days Ago.

Posted 29 April 2010 - 12:14 PM

I would be very grateful if some could do this small test, maybe we found another useful setting for our beautiful hobby http://forum.avsim.net/public/style_emot...

Kosta, I'm happy to report that (concerning the performance impact) when using LIMITED frames inside FSX, this FFTF setting of 0 completely negates the effect. So, you onto something here. (I also assumed 0 meant 'default' but setting FFTF to 0.33 impacts the frames considerably, so 0 is not equivalent to 0.33Unfortunately, (but I need to test that aspect) when i've tried this in the past the blurries were simply unnaceptable, I don't see why that would change now.Whats interesting about your discovery, is the 'impact' FFTF has on the sim when frames are limited INSIDE FSX. we are talking 6-7 FPS under heavy scenery/clouds, which doesn't make sense considering fibers run on CORE0, however it appears something 'changes' when you limit frames inside FSX. probably FIBERS run along the main scheduler on this case? In any case, I find this very interesting, specially if we can control the blurries, congrats on this discovery :)
******* (Hey-sus) &nbsp;&nbsp; Posted Image

Automated Tweaking & Tuning tool for FSX here &nbsp;|&nbsp; Download the Shader 3 Mod for FSX (w/ autoinstaller) here &nbsp;|&nbsp; My other add-ons can be found here

Intel i7 920 @ 4.2Ghz - 6GB 7-7-7-16 @ 1603, GIGABYTE GTX 480 @ 800/1600/1900 Win7 Ultimate 64-Bit w/FSX SP2, REX2, GEX, UTX-USA, ORBX PNW, FSGenesis Mesh + USA Landclass, ASE, FSPassengers, Level-D 767 and SquawkBox + Ultimate Traffic 2 (SP1) SMS File (100% VATSIM flyer in VC w/TrackIR4)

#12 bojote

bojote

    Member - 1,000+

  • Members
  • 1,565 posts
  • Joined 12 Years, 4 Months and 28 Days Ago.

Posted 29 April 2010 - 12:33 PM

In any case, I find this very interesting, specially if we can control the blurries, congrats on this discovery :)

In fact... FFTF set to 0.01 does NOT negate the effect and introduces HORRIBLE blurries, setting it to 0, NEGATES the performance impact and I don't see any blurries, Kosta, could be a setting we both have in common that is causing this behavior????
******* (Hey-sus) &nbsp;&nbsp; Posted Image

Automated Tweaking & Tuning tool for FSX here &nbsp;|&nbsp; Download the Shader 3 Mod for FSX (w/ autoinstaller) here &nbsp;|&nbsp; My other add-ons can be found here

Intel i7 920 @ 4.2Ghz - 6GB 7-7-7-16 @ 1603, GIGABYTE GTX 480 @ 800/1600/1900 Win7 Ultimate 64-Bit w/FSX SP2, REX2, GEX, UTX-USA, ORBX PNW, FSGenesis Mesh + USA Landclass, ASE, FSPassengers, Level-D 767 and SquawkBox + Ultimate Traffic 2 (SP1) SMS File (100% VATSIM flyer in VC w/TrackIR4)

#13 Kosta

Kosta

    Member - 8,000+

  • Topic Starter
  • Members
  • 9,086 posts
  • Joined 11 Years and 17 Days Ago.

Posted 29 April 2010 - 12:59 PM

KostaI have Tested your theory as you set out above, and Im happy to report you are correct about this.I have taken this one step further, and tested with my saved default test set up: JS41, FSDT JFK, Cloud cover, and fully tweaked .cfg. The result was exactly the same.adding FIBER_FRAME_TIME_FRACTION=0 frames are locked at 30fps, and we are now able to achieve 30fps external and 25-30fps in the VC.......Well done my friend...!!EDIT: hmmmm... there maybe some texture issues present with this tweak in, some of the ground scenery & textures are not being rendered correctly, will do some more testing

The tweak might be heavily dependant on all settings, especially affinity mask.

Haven't tried it (yet) but I know that FFTF lower than the default 0.33 gives blurries on my system: with FFTF at zero no time is allocated at all to load texture. It seems to me that this test may be nice for getting good fps while standing on an airfield, but it looks rather useless for actual flying...? What exactly do you hope to achieve with this test? Is it just a test to check out if fps goes back to (in your case) 73 or should it help us with getting FSX to run even better?

Jeroen, I tested this with flight of course. Not only on static field. It has proven here that textures load very quickly no matter what I do. I tried really hard panning, PMDG 747, JS41, photoscenery, around heavy scenery, lots of AG, really many tests... not one blurry texture. My hope is to achieve no FPS drop with even slightly limited FPS (99, max limit in FSX).I started very simple: I was at my home field in 747 cockpit, and I noticed with FPS Limiter, I was getting 30fps, and suddenely with internal limiter 20fps. That was annoying, especially when with external FPS Limiter, I have very big trouble of textures slowly loading and eventually causing slight blurries. If I pressed Pause, highres textures would actually load. But that is simply too late. And with this tweak, I achieve 30fps with internal limiter, with which textures also load as they do with all other usual settings.I believe that FFTF is only for the first core, Core0 on my system, which does only fibers, and doesn't affect the main scheduler as it's offloaded. I think FFTF=0 gives FSX the needed balance for the Core0 to use it fully for the fibers. I dunno, but that's my only explanation.

After further testing my reports of texture problems were misplaced, I cannot reproduce the issue again.I am also not seeing any more or less texture bluring, with or without this tweak added.What I am seeing now, is an incredibly smooth sim locked at 30fps with NO drop off in performance.... http://forum.avsim.net/public/style_emot...

Glad to hear it!!!

Hello,There had been considerable discussion on FFTF with ACES (Phil Taylor) and of course many in the community. The link below may be something you wish to read. At the very least, it provides a little more information to help you decide. I personally do not use it but again like so many tweaks, "your mileage may vary"Also please note that if you have a really speedy well tuned machine, you may not notice any change from min to max. What many tweaks have in common is that if a value is set that is lower than or greater than the allowed value, FSX will set this to a default level after the variable is read into the FSX executable. I have no information as to what happens with this value internally when it is set to zero. I is my understanding that this tweak has the greatest affect on single core machines.http://blogs.msdn.co...f-the-week.aspxRegardsBob G

I know this thread, and read it more than once, thank you. And one can notice he says that FFTF affects ONLY first core. Only the one running fibers. So the 0 tweak is actually coming to place. Since 0 setting tells it to fully run fibers. Kinda like telling it, not to use it for anything else... and it's working here!

#14 Kosta

Kosta

    Member - 8,000+

  • Topic Starter
  • Members
  • 9,086 posts
  • Joined 11 Years and 17 Days Ago.

Posted 29 April 2010 - 01:07 PM

Kosta, I'm happy to report that (concerning the performance impact) when using LIMITED frames inside FSX, this FFTF setting of 0 completely negates the effect. So, you onto something here. (I also assumed 0 meant 'default' but setting FFTF to 0.33 impacts the frames considerably, so 0 is not equivalent to 0.33Unfortunately, (but I need to test that aspect) when i've tried this in the past the blurries were simply unnaceptable, I don't see why that would change now.Whats interesting about your discovery, is the 'impact' FFTF has on the sim when frames are limited INSIDE FSX. we are talking 6-7 FPS under heavy scenery/clouds, which doesn't make sense considering fibers run on CORE0, however it appears something 'changes' when you limit frames inside FSX. probably FIBERS run along the main scheduler on this case? In any case, I find this very interesting, specially if we can control the blurries, congrats on this discovery :)

*******, I never go really heavy blurries in FSX. I don't know really why. Maybe tuned system? Or I simply didn't fly long enough. When I see that when I pause the flight textures load afterwards, its enough proof for me that something ain't working.You really think that fibers might go to Core1 in that case? What would be the point... besides, Core1 is already at 100%, why would MS program even that from Core0 away... it's kinda unlogical.Maybe we should start checking disk usage with FFTF=0? Who knows, maybe disks can't keep up? My WD is doing pretty well, and as I said... heavy situations, no blurries.

In fact... FFTF set to 0.01 does NOT negate the effect and introduces HORRIBLE blurries, setting it to 0, NEGATES the performance impact and I don't see any blurries, Kosta, could be a setting we both have in common that is causing this behavior????

Hmmm... let me do some testing. I'll run with with 0.01.

#15 bojote

bojote

    Member - 1,000+

  • Members
  • 1,565 posts
  • Joined 12 Years, 4 Months and 28 Days Ago.

Posted 29 April 2010 - 01:19 PM

Hmmm... let me do some testing. I'll run with with 0.01.

forget that, I was running UNLIMITED with that test, running LIMITED you can't see any blurries.One final test:[TERRAIN]SWAP_WAIT_TIMEOUT=0the above line, removes the 'timeout' for fiber swaping, if you use, say a 0.50 value in FFTF with a SWAP_WAIT_TIMEOUT=0 and you run UNLIMITED frames you'll see black squares flashing all over the place, this happens because the fiber responsible for updating the tiles timeouts before its even finished.Running UNLIMITED, with a FFTF of 0.01 and a SWAP_WAIT_TIMEOUT=0 won't show any 'black squares' because you are allocating very few resources to the fibers, so they never have the chance of 'timing out'Yet, whats interesting is that if running LIMITED frames, and using a SWAP_WAIT_TIMEOUT=0 you DO see the black squares flashing all over the place EVEN if FFTF is set to 0! so THERE IS fiber usage going on, even WITH FFTF=0Unfortunately, there is no logic behind all this, but the results are very promising, great find!
******* (Hey-sus) &nbsp;&nbsp; Posted Image

Automated Tweaking & Tuning tool for FSX here &nbsp;|&nbsp; Download the Shader 3 Mod for FSX (w/ autoinstaller) here &nbsp;|&nbsp; My other add-ons can be found here

Intel i7 920 @ 4.2Ghz - 6GB 7-7-7-16 @ 1603, GIGABYTE GTX 480 @ 800/1600/1900 Win7 Ultimate 64-Bit w/FSX SP2, REX2, GEX, UTX-USA, ORBX PNW, FSGenesis Mesh + USA Landclass, ASE, FSPassengers, Level-D 767 and SquawkBox + Ultimate Traffic 2 (SP1) SMS File (100% VATSIM flyer in VC w/TrackIR4)