Horrible FPS

[Fiorentoni 4,068]

10 hours ago, GCBraun said:

Ah, ok. Thought that would be a bios change or something similar. 

No it can be really easy and without any effect on the rest of the system / other programs. You basically only have to do two little things: 1. (only if HT is on) mask the three cores P3D uses, so P3D doesn't interfere with itself (that's a longer explanation, but not necessary to understand), 2. keep all third party software (ActiveSky, AIG etc.) off the cores used by P3D. Number 1 is being done by simple edit of the CFG, number 2 can be done either by Process Lasso or more simple by a batch file.
If you do that, your FPS will become much more stable. That doesn't mean you will have "more" FPS, but it means there is much less temporary FPS-loss when panning/taking off/approaching (called "stuttering"). The explanation - again - is a bit longer, but basically stuttering is being produced if P3D's main core ( the one it uses for almost anything) is being used by other processes (this happens only temporarily, so it's a short, small stutter).
I will guarantee you that it's worth it in regards to the immersion feeling. I can help you, if you want to do this.

Edited January 1, 2022 by Fiorentoni

[lownslo 590]

14 minutes ago, Fiorentoni said:

No it can be really easy and without any effect on the rest of the system / other programs. You basically only have to do two little things: 1. (only if HT is on) mask the three cores P3D uses, so P3D doesn't interfere with itself (that's a longer explanation, but not necessary to understand), 2. keep all third party software (ActiveSky, AIG etc.) off the cores used by P3D. Number 1 is being done by simple edit of the CFG, number 2 can be done either by Process Lasso or more simple by a batch file.
If you do that, your FPS will become much more stable. That doesn't mean you will have "more" FPS, but it means there is much less temporary FPS-loss when panning/taking off/approaching (called "stuttering"). The explanation - again - is a bit longer, but basically stuttering is being produced if P3D's main core ( the one it uses for almost anything) is being used by other processes (this happens only temporarily, so it's a short, small stutter).

This!! (mostly 😁).  It's not all that complicated... set the Affinity Mask of P3D (which must NOT include the Cores/Logical Processors used by addons and OS), then set the P3D Core Affinity, and then set the Schedulers.  None of this requires 3rd party apps (like Process Lasso)... LM has now given us all the tools we need in the [Job Scheduler] section.  The  three Schedulers must not include a second LP (if using HT).  Do this and the sim is very smooth, and a consistent FPS.  And if you still get CTD's after doing this, you have more issues than a simple Affinity Mask FUBAR!

[GCBraun 2,390]

So, I have done some tweaks on the CPU priority with Process Lasso, shifting Chaseplane, Activesky and some other add-ons to cores 8-19. With this, I was able to reduce Core 0 usage to 75-85%, instead of 100%.

Not sure if this helps much, but I've noticed that the overall FPS seem to be more stable.

Now I would like to proceed with the CFG changes. Is there a clear guide on how to do this based on CPU model and HT configuration?

Hopefully these tweaks will help, but I would still hope for more to be done in LM´s side to better use the capabilities of our modern GPUs. 

Just for comparison, here is some data of my GPU usage during cruise in P3D and XP11:

P3D

XP11

Edited January 1, 2022 by GCBraun

[Fiorentoni 4,068]

33 minutes ago, GCBraun said:

So, I have done some tweaks on the CPU priority with Process Lasso, shifting Chaseplane, Activesky and some other add-ons to cores 8-19. With this, I was able to reduce Core 0 usage to 75-85%, instead of 100%.

Not sure if this helps much, but I've noticed that the overall FPS seem to be more stable.

Now I would like to proceed with the CFG changes. Is there a clear guide on how to do this based on CPU model and HT configuration?

Hopefully these tweaks will help, but I would still hope for more to be done in LM´s side to better use the capabilities of our modern GPUs. 

Just for comparison, here is some data of my GPU usage during cruise in P3D and XP11:

P3D

XP11

IF you have HT on, then you have 20 logical cores (with 10 physical ones), so here's what you gotta do. Just copy that part and replace the appropriate lines in the cfg:


[JobScheduler]
AffinityMask=1048533
P3DCoreAffinityMask=1048533
MainThreadScheduler=0
RenderThreadScheduler=1
FrameWorkerThreadScheduler=2

[GCBraun 2,390]

23 minutes ago, Fiorentoni said:

IF you have HT on, then you have 20 logical cores (with 10 physical ones), so here's what you gotta do. Just copy that part and replace the appropriate lines in the cfg:


[JobScheduler]
AffinityMask=1048533
P3DCoreAffinityMask=1048533
MainThreadScheduler=0
RenderThreadScheduler=1
FrameWorkerThreadScheduler=2

Thank you so much. I was on cruise, saved the flight and changed the CFG file. Upon re-loading, I still see considerable frame instability, as seen by the 99% FPS indication of the Nvidia overlay:

Nevertheless, I will give it a bit more time to draw further conclusions.

[PinkPony 171]

On 12/27/2021 at 1:39 PM, SKEWR said:

I flew with Enhanced Atmospherics off and there were no issues. I'll continue on other test flights and report back if I have any horrible FPS without EA on.

SKEWR,

Since setting EA OFF have you entirely solved the sudden/permanent FPS drop/freeze?

How many sessions so far?

I have completed a total of four sessions in v5.3 HF1.1 Three using FSL A320SL/321SL and one using the PMDG B748.

v5.1 HF1.1 with EA OFF
 

Previously in v5.3 using EA ON, 25% of sessions (40+ sessions flown) encounter the bug.

 

Edited January 1, 2022 by PinkPony

[virtualstuff 750]

17 hours ago, PinkPony said:

SKEWR,

Since setting EA OFF have you entirely solved the sudden/permanent FPS drop/freeze?

How many sessions so far?

I have completed a total of four sessions in v5.3 HF1.1 Three using FSL A320SL/321SL and one using the PMDG B748.

v5.1 HF1.1 with EA OFF
 

Previously in v5.3 using EA ON, 25% of sessions (40+ sessions flown) encounter the bug.

 

Go to an Airport with dynamic lights and start playing with the time from morning till night and after several slider movements forward backward in time you will have your bug 😉

[SKEWR 364]

On 1/1/2022 at 3:01 PM, PinkPony said:

SKEWR,

Since setting EA OFF have you entirely solved the sudden/permanent FPS drop/freeze?

How many sessions so far?

I have completed a total of four sessions in v5.3 HF1.1 Three using FSL A320SL/321SL and one using the PMDG B748.

v5.1 HF1.1 with EA OFF
 

Previously in v5.3 using EA ON, 25% of sessions (40+ sessions flown) encounter the bug.

Hiya Pink Pony!!!

I've done 6 flights in the same area (China. Korea, Japan) and my stutters have not come back. I CAN have EA on but NOT volumetric clouds. I've been using the PMDG 747-8 for all flights.

Have you had on volumetric clouds?

[PinkPony 171]

  

4 minutes ago, SKEWR said:

Hiya Pink Pony!!!

I've done 6 flights in the same area (China. Korea, Japan) and my stutters have not come back. I CAN have EA on but NOT volumetric clouds. I've been using the PMDG 747-8 for all flights.

Have you had on volumetric clouds?

EA OFF since upgrading to v5.3 HF1.1 - Four problem free sessions, FPS drop/freeze just occurred during session number five (see below). 

Upon requesting pushback via GSX (FSL A321SL at FlyTampa CYYZ) the sudden/permanent FPS/DROP freeze occurred again. Prior to that, 30 FPS via VSYNC (30Hz monitor). 

As @B777ER confirmed, I can now also confirm that EA ON/OFF makes no difference (if anything only how frequently the bug occurs).

As a final attempt prior to giving up on v5.3 HF1.1 for good (and returning to v5.1 HF1) I have downgraded my Nvidia driver (RTX 3080Ti) from 496.09 to 471.96. 

After going through hundreds of posts from users suffering from this bug, I noted that two users who claim they have yet to experience this bug in v5.1 HF1.1 have been using 472.12 (or older) drivers. It is no secret that the Nvidia 496+ driver branch has introduced numerous significant bugs. Possibly something causing this bug in v5.3/HF/HF1.1 ?

During this most recent occurrence of the sudden/permanent FPS drop/freeze, I noticed t FPS rapidly alternating between 0 FPS and 60 FPS. I always run at 30 FPS via VSYNC (30Hz monitor) so it appears that VSYNC no longer functions when the bug occurs. Unsure if that is the cause or merely a consequence of what is occurring within the v5.3/HF1/HF1.1 application.
 

Another five sessions coming up with 471.96 to try and confirm if the Nvidia driver could possibly be the a culprit and the solution to this bug. 

It is a word not allowed shame that v5.3 is plagued by this bug. The majority of sessions perform flawlessly with the best P3D performance to date. LM needs to figure this out ASAP. 

Edited January 3, 2022 by PinkPony

[mwa05 108]

23 minutes ago, PinkPony said:

As a final attempt prior to giving up on v5.3 HF1.1 for good (and returning to v5.1 HF1) I have downgraded my Nvidia driver (RTX 3080Ti) from 496.09 to 471.96. 

I use studio drivers only and am on the latest 472.84 (3080Ti) and I quite literally have flown all around the world over the last week or so with 5.3 HF1+1 using the PMDG 777 and FSLabs A320 almost daring the issue to occur including real time long haul SYD-SIN and SIN-ZRH (777) and DOH-HEL (A320) and nothing.

I use EA On, VOL Off, with ASP3D 8027 with ASCA and the TOGA products with GSX, and I have all Orbx Global Products.

I would not normally come onto a thread to state I don’t have a problem as I know that is annoying to those who do and are trying to solve it (a bit like saying my car is stolen and someone else saying well my car is not stolen - it takes you nowhere!) but I can confirm no issues on Nvidia Studio Drivers, and can back that in with more than 40 hours of real time 5.3 HF1+1 flying (as I am home in an empty house as my wife and children are overseas visiting her mother who due COVID they have not seen for 2 years). 

So I have been able to have a sim-a-thon.

Mark

[PinkPony 171]

19 hours ago, mwa05 said:

I use studio drivers only and am on the latest 472.84 (3080Ti) and I quite literally have flown all around the world over the last week or so with 5.3 HF1+1 using the PMDG 777 and FSLabs A320 almost daring the issue to occur including real time long haul SYD-SIN and SIN-ZRH (777) and DOH-HEL (A320) and nothing.

I use EA On, VOL Off, with ASP3D 8027 with ASCA and the TOGA products with GSX, and I have all Orbx Global Products.

I would not normally come onto a thread to state I don’t have a problem as I know that is annoying to those who do and are trying to solve it (a bit like saying my car is stolen and someone else saying well my car is not stolen - it takes you nowhere!) but I can confirm no issues on Nvidia Studio Drivers, and can back that in with more than 40 hours of real time 5.3 HF1+1 flying (as I am home in an empty house as my wife and children are overseas visiting her mother who due COVID they have not seen for 2 years). 

So I have been able to have a sim-a-thon.

Mark

Mark, thank you for sharing!

I am currently conducting my second session with Nvidia 471.96 (the first four hour long flight yesterday was perfect with the FSL A321).

However, due to this bug only appearing suddenly and sporadically (~1/4 sessions) I will need to perform at least 5 more sessions to verify that the Nvidia 497 series driver are to blame for this bug in V5.3/HF/HF1.1 (I sure hope it is). 
 

Fingers crossed! 

Edited January 3, 2022 by PinkPony

[Ray Proudfoot 8,903]

20 minutes ago, PinkPony said:

I am currently conducting my second session with Nvidia 471.96

I’m still on those drivers to help combat the DXGI problem. Together with the Registry fix I’ve had no repeat of the problem for over two months. I’m loath to move away from them.

[PinkPony 171]

2 hours ago, Rob_Ainscough said:

I was able to replicate horrible FPS and long pauses at MK Studio's EFHK (just sitting on the ground in a default F35) and have passed along my logging data to LM and hopefully they can see what might be happening.  MK Studios did an incredible job with the airport (visually stunning) but clearly P3D V5.3 seems to have some issue with it ... hopefully LM can replicate and/or get sources to ID from MK Studios.

Cheers, Rob.

Rob,

MK’s EHFK became a stutterfest in P3D v5.2HF1 

https://forum.orbxdirect.com/topic/208690-efhk-stutters-mk-studios/

 

Unfortunately I have just encountered the FPS drop/freeze during my second session using Nvidia driver 471.96 

Unsure what else I can possibly try at this point. LM needs to sort this out.

Back to V5.1 HF1 for now

[PinkPony 171]

33 minutes ago, Rob_Ainscough said:

Well maybe MS EHFK will give LM a clue or a starting point because as of right now they haven't been successful at recreating the issues you folks have reported.  So far, EHFK is the only place I've been able to generate pauses (more than just stutters).

Cheers, Rob.

I agree and I hope that the two are related!

I must say I am both shocked and disappointed if LM really have failed to reproduce this after weeks of bug reports.

Myself and dozens of other experienced P3D users can reproduce this consistently yet LM haven’t a single time?

Not to mention, the FSlabs dev team have already confirmed this bug in their own testing of V5.3 and V5.3HF1 and have downgraded back to V5.2HF1 as a result…

Can’t be that difficult for LM to reach out (or vice versa). I’m sure the FSLabs devs can share insight form their own debugging. 

215+ posts on the main bug thread over on the P3D forums. Numerous users detailing how they are consistently able to reproduce this and outlining steps to do so. Yet LM can’t reproduce a single time?

Blackbox711 (a highly respected P3D user and Twitch streamer) experienced this bug on multiple occasions during livestreams using V5.3

Not to mention P3D Core Lead Rob McCarthy recently stating that they implemented some fixes in HF to try and solve this.

Sounds more like they know very well what is going on but simply don’t know what is causing it or how to fix it yet.

Edited January 4, 2022 by PinkPony

[Flyinggok 55]

Just an idea, probably we should start telling LM at which location/airport we seeing this issue.

Also if its starting during flight, which airport/scenery we are currently overflying.

I can only guess, but it looks more and more like a scenery / airport problem that some scenery are causing this issue for whatever reason.
I hope not, because I have little hope that some devs will adapt their scenery to a new P3D version.

I saw the issue at Simwings Munich, with Orbx Base, OpenLC, Germany South installed. FSLabs was used all time.

• 3 times at the Gate during boarding
  • 1 time it recoverd but started than again during taxi.
• 2 times during take off and taxi
• 1 time it didt not occur

 

 

Edited January 4, 2022 by Flyinggok