FFPS:FFTF question

[pmb 5,137]

Hi, thanks, actually I AM quite please with what I have now. There's (rather rarely) autogen loading in patches, sometimes terrain snapping in place (which I doubt the tool will help with), but I see very rarely blurries (given I mostly flow slow GA planes). Plus a bit jitter when fps go below 30 Hz in very hard environments. 

No need to hurry up, at least 

Kind regards, Michael

[joepoway 241]

30 minutes ago, pmb said:

Hi, thanks, actually I AM quite please with what I have now. There's (rather rarely) autogen loading in patches, sometimes terrain snapping in place (which I doubt the tool will help with), but I see very rarely blurries (given I mostly flow slow GA planes). Plus a bit jitter when fps go below 30 Hz in very hard environments. 

No need to hurry up, at least 

Kind regards, Michael

Michael

I'm in the exact same boat as you. I played a bit today with just adding FFTF values and leaving unlimited, TB on and Vsync with my 30Hz monitor. I'm not sure it helped much I tried 0.05, 0.08, 0.1, 0.15, 0.2 it may have helped with my occasional blurries, the good thing was I didn't see any FPS drop or stutters until 0.2 with a PMDG aircraft and even then not much. On a different note I also increased my LOD from high (12km) to Max (30km) because I believe I have margin but it was also marginal. So for now I'm going to fly with an FFTF= 0.1 for a bit with the LOD at Max and see if I notice any difference.

The thing that bothers me the most is the late autogen and the airports like Flightbeam KSFO going from black foot prints to fully populated when approaching from the distance or panning around in external views out a few miles. I'm not sure any of this will help either of those things. I saw all your posts on the P3D forum back several months ago but LM didn't have an answer.

One other thing I notice your CPU runs at 4.0GHz have you tried to OC that a bit?

Best Regards

Joe

Edited March 24, 2018 by joepoway

[SteveW 1,792]

What we see is LM continuing to push. Keeping up the work, sitting down with the cup of coffee and getting on with it every day. It's a long road and it's going to get better as they tune here and there to keep things in line with current hardware and so on. Can things be improved on their part with respect to allocating resources more efficiently? We shall see, though there has been plenty so far. What we see is a lot of planes and scenery but we might think the under-the-hood tech is creeping along perhaps. It's not slow, it's just because it's tricky.

[orbmis 9]

20 minutes ago, joepoway said:

Michael

I'm in the exact same boat as you. I played a bit today with just adding FFTF values and leaving unlimited, TB on and Vsync with my 30Hz monitor. I'm not sure it helped much I tried 0.05, 0.08, 0.1, 0.15, 0.2 it may have helped with my occasional blurries, the good thing was I didn't see any FPS drop or stutters. On a different note I also increased my LOD from high (12km) to Max (30km) and it was also marginal. So for now I'm going to fly with an FFTF= 0.1 for a bit with the LOD at Max and see if I notice any difference.

 

I’m also in the same situation as Joe,,, various changes to FFTF, all the way up to .60 . FPS holds steadily at 60fps (Vsync/Tb w/unlimited), no stutters, but marginal, if any, improvements to visuals/blurries. I’m starting to think my CPU may be the root of my issues, but I don’t see the CPU being pressured when in flight. Maybe my GTX1060 SC 6gb just cant handle the load?

mike

Edited March 24, 2018 by orbmis

[pmb 5,137]

Thanks guys for the experiences, much appreciated. I tried FFTF = 0.01 some time ago (nothing changed elsewhere), but didn't notice much change. Thus I thought I'd better leave out any more tweak which didn't help anyway. I might do some more experiments one day.

The black buildings are ugly indeed. Fly from KSFO into the Silicon Valley and Moffett Field will come up black. Fly from Niagara Falls over the Ontario Lake and the Skyline of Toronto will show in black, too. I doubt the FFTF tool will help here. If anyone reports it to do I'll buy it on a whim.

I never overclocked my CPUs. I've to do real work on the machine, too, and I can't compromise reliabiliy in any way. After 4 years my wife takes my machines for another 4 years, so lifetime does matter. Besides, I can run HT on, which I found to improve texture loading noticeably on my machine (contrary to former P3D versions).

Kind regards, Michael

[Cruachan 705]

Hi,

Well, this is a little awkward as I had started out believing that this utility might be the answer to all our prayers. After spending a whole afternoon and evening exhaustively testing FFTF Dynamic, along with my search for the optimum Affinity Mask, I regret to report that ‘manna from heaven, it isn’t. Of course, my experience is confined to how Prepar3D performs under the good auspices of my trusty 8 cored i7-5960X with hyperthreading enabled. So, my results may well be totally unrepresentative of what can be achieved with 4 or even 6 cored CPUs.

It should be noted that my rig is not dedicated solely for the use of Prepar3D so I choose to keep Hyperthreading enabled.

I should remind everyone that my hope was that introducing this utility would eliminate the need to switch to a less demanding profile while visiting frame rate hogging situations in our simming world. Also, as described in a previous post, I was experiencing intermittent ‘juddering’ which, at times could be quite persistent and annoying and I was keen to determine whether this was being caused by unrealistic settings or a competing external process unrelated to the running sim.

KSEA was chosen as my testing area as it is known to be a very demanding situation with a high density of scenery objects. I started from scratch with a fresh Prepar3D.cfg, cleared the shaders folder and reset the Prepar3D entries in NVidia Inspector to their defaults. FXAA was turned off and I chose to use 8xMSAA in-sim (instead of the previously employed 4xSGSSAA in NI along with 4xMSAA in Prepar3D). Otherwise, the initial default in-sim settings were left unchanged and no AM was added to the configuration file. Predictably performance was very good with relatively high frame rates (Frames Unlimited) preserved throughout several test flights. There were a few micro stutters here and there, but they were brief and reminiscent of my experiences with P3Dv3.4.

Settings were slowly increased until the sim was struggling on the ground at KSEA. FFTF Dynamic was then tried using both modes in turn and, while this seemed to be helpful, fluidity was compromised with the appearance of frequent stuttering. Attempts to mitigate this were made but nothing seemed to help much. The stuttering persisted and so FFT Dynamic was closed and set aside for now.

My attention then turned to the Affinity Mask (AM). Steve’s suggestions were followed and it quickly became apparent that the Prepar3D Main thread did not ‘like’ being moved away from Core 0 on my chip.The sim began stuttering from the word go and these were very persistent to the extent that it soon became unflyable. In addition, blurries were everywhere and I noted that all the threads running on the LPs contained within the Mask, other than the main thread, were maxing out. I’ve seen this behaviour before and it coincided with the previously mentioned episodic juddering. Also, this was happening with fairly conservative in-sim Settings. Testing following warm and cold rebooting did not help. This was alarming to the extent that I was starting to suspect that this might be a hardware issue.

Testing continued using many different permutations (including various frame locked values, Unlimited with and without VSync and TB) for a while thereafter, but it was clear that I was getting nowhere. That only the ace up my sleeve which was known to deliver the performance I desired: I reintroduced my AM value of 21845 (01,01,01,01,01,01,01,01) and all was well, once again! The relief I felt at that moment was quite palpable..lol!

Perhaps I should mention that I did try AM=21844 (01,01,01,01,01,01,01,00) but sadly, again, intermittent micro stuttering appeared. This whole thing is crazy and temptingly illogical, but there it is, that has been my experience..

This time things are different. Those juddering episodes seem to have gone or, at least are hardly noticeable and very transient in nature. Threaded activity appears more normal and only max out with brief spikes of LP activity. My 1080Ti never seemed stressed at any point throughout this testing marathon.This after reinstating my preferred sensible settings in-sim. I have to conclude, therefore, that this may have had something to do with my use of SGSSAA. Currently 8xMSAA seems to be doing a pretty good job and a flight through the Rockies (ORBX) produced some wonderful sharp mountain textures as they slowly drifted by my Real Air Duke Turbine V2.

Hope all this makes some sense.

Mike

Edited March 24, 2018 by Cruachan
Rephrasing and punctuation corrections.

[Ted Striker 338]

21 minutes ago, Cruachan said:

That only left my ace card which was known to deliver the performance I desired: I reintroduced my AM value of 21845 (01,01,01,01,01,01,01,01) and all was well, once again! The relief I felt at that moment was quite palpable..lol!

Thanks for sharing the very thorough summary of your testing Mike. With that AM P3Dv4 is running on just one processor of each core. Am I missing something or does this indicate that on your machine P3Dv4 refers HT (hyperthreading) = off?

Ted

[Cruachan 705]

13 minutes ago, Ted Striker said:

Thanks for sharing the very thorough summary of your testing Mike. With that AM P3Dv4 is running on just one processor of each core. Am I missing something or does this indicate that on your machine P3Dv4 refers HT (hyperthreading) = off?

Hi Ted,

Hitherto with the Prepar3D.x series HT=OFF was the preferred configuration on my setup. As I have shown, Prepar3D v4.x still seems to prefer to run on 8 rather than 16LPs. So, this nifty solution gives me the best of both worlds. As I stated, my rig is not dedicated solely for the use of P3D although it is used extensively for that purpose. Everything else including Aerofly FS2, FS World, X-Plane 11 and several demanding STEAM sourced games appear to get along fine with HT enabled. I have not encountered any heat issues so I decided to leave it at that. Such flexibility is both useful and welcome.

Regards,

Mike

[Nemo 1,305]

1 hour ago, Cruachan said:

Hi,

Well, this is a little awkward as I had started out believing that this utility might be the answer to all our prayers. After spending a whole afternoon and evening exhaustively testing FFTF Dynamic, along with my search for the optimum Affinity Mask, I regret to report that ‘manna from heaven, it isn’t. Of course, my experience is confined to how Prepar3D performs under the good auspices of my trusty 8 cored i7-5960X with hyperthreading enabled. So, my results may well be totally unrepresentative of what can be achieved with 4 or even 6 cored CPUs.

It should be noted that my rig is not dedicated solely for the use of Prepar3D so I choose to keep Hyperthreading enabled.

I should remind everyone that my hope was that introducing this utility would eliminate the need to switch to a less demanding profile while visiting frame rate hogging situations in our simming world. Also, as described in a previous post, I was experiencing intermittent ‘juddering’ which, at times could be quite persistent and annoying and I was keen to determine whether this was being caused by unrealistic settings or a competing external process unrelated to the running sim.

KSEA was chosen as my testing area as it is known to be a very demanding situation with a high density of scenery objects. I started from scratch with a fresh Prepar3D.cfg, cleared the shaders folder and reset the Prepar3D entries in NVidia Inspector to their defaults. FXAA was turned off and I chose to use 8xMSAA in-sim (instead of the previously employed 4xSGSSAA in NI along with 4xMSAA in Prepar3D). Otherwise, the initial default in-sim settings were left unchanged and no AM was added to the configuration file. Predictably performance was very good with relatively high frame rates (Frames Unlimited) preserved throughout several test flights. There were a few micro stutters here and there, but they were brief and reminiscent of my experiences with P3Dv3.4.

Settings were slowly increased until the sim was struggling on the ground at KSEA. FFTF Dynamic was then tried using both modes in turn and, while this seemed to be helpful, fluidity was compromised with the appearance of frequent stuttering. Attempts to mitigate this were made but nothing seemed to help much. The stuttering persisted and so FFT Dynamic was closed and set aside for now.

My attention then turned to the Affinity Mask (AM). Steve’s suggestions were followed and it quickly became apparent that the Prepar3D Main thread did not ‘like’ being moved away from Core 0 on my chip.The sim began stuttering from the word go and these were very persistent to the extent that it soon became unflyable. In addition, blurries were everywhere and I noted that all the threads running on the LPs contained within the Mask, other than the main thread, were maxing out. I’ve seen this behaviour before and it coincided with the previously mentioned episodic juddering. Also, this was happening with fairly conservative in-sim Settings. Testing following warm and cold rebooting did not help. This was alarming to the extent that I was starting to suspect that this might be a hardware issue.

Testing continued using many different permutations (including various frame locked values, Unlimited with and without VSync and TB) for a while thereafter, but it was clear that I was getting nowhere. That only the ace up my sleeve which was known to deliver the performance I desired: I reintroduced my AM value of 21845 (01,01,01,01,01,01,01,01) and all was well, once again! The relief I felt at that moment was quite palpable..lol!

Perhaps I should mention that I did try AM=21844 (01,01,01,01,01,01,01,00) but sadly, again, intermittent micro stuttering appeared. This whole thing is crazy and temptingly illogical, but there it is, that has been my experience..

This time things are different. Those juddering episodes seem to have gone or, at least are hardly noticeable and very transient in nature. Threaded activity appears more normal and only max out with brief spikes of LP activity. My 1080Ti never seemed stressed at any point throughout this testing marathon.This after reinstating my preferred sensible settings in-sim. I have to conclude, therefore, that this may have had something to do with my use of SGSSAA. Currently 8xMSAA seems to be doing a pretty good job and a flight through the Rockies (ORBX) produced some wonderful sharp mountain textures as they slowly drifted by my Real Air Duke Turbine V2.

Hope all this makes some sense.

Mike

Mike, thank you for this critical review. As I decided before, I will stay away from this add-on. But as you and others said, dynamic FFTF might be more useful for 4 and 6-core machines. I will also try your setting with 8 core HT on (16 LP) together with AM (01,01,01,01,01,01,01,01). Until now I have only tried HT off or HT on (but no AM setting).

Edited March 24, 2018 by Nemo

[Cruachan 705]

Hi,

For those who might be interested, what follows are the images of my current in-sim settings and what is achieved (frame rate wise - see bottom right hand corner of first image) at the KVPS) default situation with Fair Weather:

Note: The first image was captured with the Target frame rate set temporarily to Unlimited (without VSync and TB). My preferred setting these days is locked at 30 although I will switch to Unlimited as the occasion may arise (e.g. a detailed airport), or simply switch to a less demanding Profile. It's truly amazing how good the sim looks with quite conservative settings.

The sim's lighting is courtesy of Adam's 26_02 Preset for PTA 2.61

In this instance, the relevant addons include ORBX Global BASE, and OpenLC North America, FSGlobal Ultimate NG (Little affect as area very flat for as far as the eye can see), Black Marble BASE and VECTOR, MyTraffic Pro 6.0a, Probably forgotten a few but, in any case, they are unlikely to have much in the way of additional impact.

I have often wished I could upload a video to demonstrate the sim's performance, but as long as my upload speeds remain sub 1 Mb/sec that's unlikely to happen - thanks BT!

Regards,

Mike

 

[Guest]

AS far as I can tell having made some comparisons FFTF does nothing if you run P3D V4 with Unlimited FPS in the display setting. If you lock your Frame Rate in Prepar3D V4 FFTF does make a difference. I say 24fps at FFTF=0.1 and 13 FPS at FFTF=0.8 on the ground at KLAX FSDT. With unlimited Frames I saw 24fps with FFTF=0.1 and 24fps with FFTF=0.8.

I suspect that FFTF does nothing at Unlimited FR so this app is not for me.

[joepoway 241]

7 hours ago, pmb said:

Thanks guys for the experiences, much appreciated. I tried FFTF = 0.01 some time ago (nothing changed elsewhere), but didn't notice much change. Thus I thought I'd better leave out any more tweak which didn't help anyway. I might do some more experiments one day.

Michael

Just a note since you run unlimited your FFTF value is defaulted to 0.01 as I understand it, that’s why I have been experimenting with increasing it manually a little bit to see if the dynamic tool maybe worth the go.

Joe

Edited March 24, 2018 by joepoway

[pmb 5,137]

4 minutes ago, joepoway said:

Michael

Just a note since you run unlimited your FFTF value is defaulted to 0.01 as I understand it, that’s why I have been experimenting with increasing it manually a little bit to see if the dynamic tool maybe worth the go.

Joe

Thanks Joe, this is in line with Avidean's answer and does make sense.

I'll probably just stay with what I use now, but follow others' experiences despite.

Kind regards, Michael

[Guest]

23 minutes ago, joepoway said:

Michael

Just a note since you run unlimited your FFTF value is defaulted to 0.01 as I understand it, that’s why I have been experimenting with increasing it manually a little bit to see if the dynamic tool maybe worth the go.

Joe

OK, So you are saying that even if I have added:

[MAIN]
FIBER_FRAME_TIME_FRACTION=0.8

To the .cfg and then run P3D V4 with unlimited frame rate the sim ignores the .cfg value and uses 0.01 anyway? In which case this app is still not for me because I am not moving to a fixed Frame Rate. That will cause more problems for me than it will solve.

Edited March 24, 2018 by Guest

[pmb 5,137]

9 minutes ago, Avidean said:

OK, So you are saying that even if I have added:

[MAIN]
FIBER_FRAME_TIME_FRACTION=0.8

To the .cfg and then run P3D V4 with unlimited frame rate the sim ignores the .cfg value and uses 0.01 anyway? In which case this app is still not for me because I am not moving to a fixed Frame Rate. That will cause more problems for me than it will solve.

As far as I recall (it was some time ago), I added just 0.01. Running unlimited. Without much result, if any.

Kind regards, Michael