MSFS 2020 Auto FPS App v0.4.2

[sonny147 579]

i need a 120hz oled 

[Vali 83]

1 hour ago, Cpt_Piett said:

TLOD heavily affects main thread, absolutely. This becomes very obvious when trying to land at Heathrow with a TLOD much higher than the CPU (main thread) can handle. The mainthread will be hammered, and there will be massive stutters. The worse scenarios is in PG-cities as PG is a "mainthread-killer". Up at cruise, we should be able to run much higher TLODs without any issues. 

So what you're saying is that you get stutters (main thread frame time spikes) with FG on vs off? Not sure if I understand. Could it have something to do with the frame gen mod? Have you tried comparing AutoFPS on vs off?  

 

So I did a full restart of PC and MSFS2020 and verified the scenarios. Sorry for the many screenshots.

1. At the gate, Frame Gen OFF in sim settings, AutoFPS ON (default values)

2. Airborne, Frame Gen still OFF, AutoFPS ON

3. Frame Gen OFF, AutoFPS OFF, no significant change

4. Frame Gen ON, AutoFPS still OFF, effect on the Main Thread visible

5. Frame Gen ON, AutoFPS On but FG inactive (focus on other window than the sim)

6. Frame Gen ON, AutoFPS ON, FG Active - strangely this time smooth (60fps FG)

I do not see a problem of VRAM though. I put TLOD Max 200, seems still smooth. I do not get it anymore why sometimes stutters get in. Seems to be after gradual decrease of TLOD back to 50 during landing, but maybe I am wrong.

Edited March 24 by Vali

[Adamski_NZ 952]

Just a small observation. I've been doing a lot of low-level VFR flights in the mountainous regions of New Zealand with AutoFPS set to all sorts of experimental configurations.

AutoFPS works brilliantly but, as others have noted, there seems to be the odd occasion when you get a period of stutters. In my case, these happened in the middle of nowhere (but at low level) - so plenty of terrain mesh info (but no object load at all). The stutters cleared after 15 secs or so (with no visible change to scenery type) so I'm guessing that the stutters were Asobo related and nothing to do with [the excellent] AutoFPS.

One can only hope that SU15 will help, but performance with AutoFPS is still amazing.

[Cpt_Piett 4,696]

Following in my own footsteps of doing the complete opposite of what Reset recommends, I tried landing the Fenix at Tivat in VFR mode. Stutter-fest on final approach! I've only been able to use VFR in very remote areas, the advantage being very high TLODs close to the ground. Hence, if there's more than a tiny village in the area, IFR is the way for airliners. 

Due to the stutters, I aborted the approach and switched to IFR mode (which of course worked very well). Video: https://www.youtube.com/watch?v=gXVOgL6EuuQ

Fun fact: After a coffee and some scones in the airport cafeteria, I proceeded to Athens, where I burned some 7 tonnes of fuel whilst spending several hours in a hold at FL360 just south of the airport. During this time I edited and rendered the abovementioned video, all without MSFS crashing. Latest SU15 build is possibly the most stable MSFS build so far. 

10 hours ago, Vali said:

So I did a full restart of PC and MSFS2020 and verified the scenarios. Sorry for the many screenshots.

Thanks for the screenshots as well as the detailed description. I studied them but not sure I understand what's going on. 

10 hours ago, Vali said:

Frame Gen ON, AutoFPS still OFF, effect on the Main Thread visible

So you get more stutters with AutoFPS off? Or do you just get mainthread frame time spikes that are visible in dev mode overlay, but no visible stutters?

10 hours ago, Vali said:

I do not get it anymore why sometimes stutters get in.

I've spent a fair bit of time chasing stutters. Sometimes they're elusive and I can't for the life of me troubleshoot them. Other times, they're obvious e.g. during final approach into heavy scenery. Or even when overflying a small town at low altitude in a GA aircraft. What we have at the moment though, thanks to AutoFPS, is leaps and bounds better than what we had pre-AutoFPS!

Edited March 25 by Cpt_Piett

[fppilot 4,375]

1 hour ago, Cpt_Piett said:

Following in my own footsteps of doing the complete opposite of what Reset recommends, I tried landing the Fenix at Tivat in VFR mode. Stutter-fest on final approach! I've only been able to use VFR in very remote areas, the advantage being very high TLODs close to the ground. Hence, if there's more than a tiny village in the area, IFR is the way for airliners. 

Due to the stutters, I aborted the approach and switched to IFR mode (which of course worked very well). Video: https://www.youtube.com/watch?v=gXVOgL6EuuQ

I have stayed quiet for the past three or four days, licking my wounds after heaping high praise.  My last flight with AutoFPS was Thursday from KBJC to KTRM.  Both  with use of relatively simplistic freeware airports.  My Wednesday flight into KBJC (Rocky Mountain Muni, Denver area) was wonderfully smooth with excellent FPS for this system (58-62 reported by AutoFPS), but on approach to that airport I ran into a stutterfest.  A bad stutterfast. Then on Thursday the same occurred on my flight from KBJC to KTRM (Thermal, Calif).  Absolutely no stutters at Rockey Mtn Muni on departure, or enroute.  But then the same stutterfest on approach to Thermal.  Bad stutterfest.  I recall switching back and forth from IFR to VFR to calm things down, but to no avail. 

So why so much of a stutterfest at an airport on approach and taxi in, but nothing like that the next morning on startup and taxi out at the same airport?  And the same results at the following airport?  I am lost on what to do to overcome, but this promises so much I feel compelled to continue trying.  I am hybrid, using an RTX 3080 with the <4 series RTX frame generation mod, and the latest AutoFPS test.  0.4.3 for the Wednesday and Thursday flights mentioned above.  Have been out of pocket since Friday so no subsequent flights yet.

Edited March 25 by fppilot

[prolixindec 17]

Let me just add my appreciation for this. First of all, installation was flawless and fast, thanks for that. Second, I love the dog icon.

I use MSFS exclusively for outside visuals; I'm using Aerowinx PSX to provide everything having to do with the aircraft (systems, navigation, cockpit, ATC, etc.) and now I'm running steady 60 fps from the gate to cruise altitude, on an old computer with an old GPU. This has let me appreciate the beauty of MSFS at altitude. It's like a new sim! 

Thanks again for your efforts, it's greatly appreciated.

 

[Cpt_Piett 4,696]

1 hour ago, prolixindec said:

Second, I love the dog icon

You and me both. Leo keeps finding his way to my desktop: 

1 hour ago, fppilot said:

So why so much of a stutterfest at an airport on approach and taxi in, but nothing like that the next morning on startup and taxi out at the same airport?  And the same results at the following airport?  I am lost on what to do to overcome, but this promises so much I feel compelled to continue trying.

Yeah, this doesn't make much sense. My stutters were 100% user-induced though, thanks to me using AutoFPS well outside of its intended usage area. Yet again. Every time I do this though, I learn something new. 

I've given up trying to make things 100% stutter-free as it seem to be an inherent part of the sim. In the video I shared above, I did get micro-stutters despite doing my very best to optimize everything, searching every little nook and cranny for solutions. MSFS is an enigma, perhaps not meant to be fully understood by us mere mortals... 🤷‍♂️

Edited March 25 by Cpt_Piett

[Reset XPDR 1,271]

3 hours ago, Cpt_Piett said:

Following in my own footsteps of doing the complete opposite of what Reset recommends, I tried landing the Fenix at Tivat in VFR mode. Stutter-fest on final approach! I've only been able to use VFR in very remote areas, the advantage being very high TLODs close to the ground. Hence, if there's more than a tiny village in the area, IFR is the way for airliners. 

You are a glutton for punishment with these scenarios you put your system through! 😄I still believe your TLOD is way too high, even with your high end system, and VFR mode struggles with adjusting your TLOD down as you get close to the ground simply because it has too much of a TLOD delta to deal with. When I had the app so it was unrestricted below 1000 ft, the large TLOD drops would cause stutters for you and now with the below 1000 ft restrictions in place, your TLOD is not low enough by the time you cross 100 ft, which also causes stutters. At least with IFR mode, it takes that huge TLOD range into account and will start marching TLOD down well before final approach but VFR mode doesn't do that and can't deal with it.

As such, my recommended solution for you is:

1. If you want to use VFR mode, you need to have a much lower TLOD Max. 200 would be ideal (which is what I use now on a similar performing CPU to yours), 300 maybe and 400 max but no more.
2. If you don't want lower TLOD Max, which I suspect you don't because you are a high TLOD junkie 😄, then resign yourself to not using VFR mode, ever!

 

[Reset XPDR 1,271]

2 hours ago, fppilot said:

I have stayed quiet for the past three or four days, licking my wounds after heaping high praise.  My last flight with AutoFPS was Thursday from KBJC to KTRM.  Both  with use of relatively simplistic freeware airports.  My Wednesday flight into KBJC (Rocky Mountain Muni, Denver area) was wonderfully smooth with excellent FPS for this system (58-62 reported by AutoFPS), but on approach to that airport I ran into a stutterfest.  A bad stutterfast. Then on Thursday the same occurred on my flight from KBJC to KTRM (Thermal, Calif).  Absolutely no stutters at Rockey Mtn Muni on departure, or enroute.  But then the same stutterfest on approach to Thermal.  Bad stutterfest.  I recall switching back and forth from IFR to VFR to calm things down, but to no avail. 

So why so much of a stutterfest at an airport on approach and taxi in, but nothing like that the next morning on startup and taxi out at the same airport?  And the same results at the following airport?  I am lost on what to do to overcome, but this promises so much I feel compelled to continue trying.  I am hybrid, using an RTX 3080 with the <4 series RTX frame generation mod, and the latest AutoFPS test.  0.4.3 for the Wednesday and Thursday flights mentioned above.  Have been out of pocket since Friday so no subsequent flights yet.

Perhaps you can post up a screenshot of your settings or the contents of your config file so we can see if anything is obviously out of kilter.

Also, are you using the GPU-Z companion app feature to show your GPU load in the app? I have found in the last few days that if I don't have GPU-Z running, I get a much smoother AutoFPS experience. It is possible that the GPU-Z interface is causing an overload of the GPU at times, external to what MSFS is asking from it, and that results in these stutters. If I can get a few other users to confirm it too eg. @Cpt_Piett, I may just disable the GPU-Z integration code as it doesn't do anything in the app other than display your GPU load anyway.

Edit: I may just push out the latest test build that I have shortly and disable GPU-Z in it so testers don't have to do anything to stop GPU-Z interaction and we'll see how that goes for everyone.

Edited March 25 by Reset XPDR

[Reset XPDR 1,271]

New Test Version 0.4.2.7-test is available here. This test build aims to address stuttering issues that some have been experiencing since 0.4.2 was released, primarily by trialing the disabling of GPU-Z companion app integration as my own testing has revealed it seems to contribute to stutters. Please let me know if anything changes on the stutter front with this test build, particularly for those who have been experiencing them previously.

Changes are:

• Changed VFR mode to start up with TLOD Min like IFR mode does, rather than with the default MSFS TLOD setting like previously.
• Streamlined resetting of TLOD Min + on the ground when switching applicable settings or display modes.
• Changed TLOD Min + to use higher extra TLOD when TLOD Max is low
• Reduced TLOD Max in non-expert mode to 2 times default TLOD instead of 3 times, to reduce chance of stutters occuring at higher TLOD Max settings.
• Removed VFR mode TLOD sliding scale reduction as it can cause TLOD to be too high on landing and seems to cause worse stutters than if it marched down much quicker on that same approach.
• Disabled GPU-Z companion app integration temporarily, perhaps permanently as the information it gets is not used in the app for anything other than displaying GPU load, as it has been shown on a few occassions to correlate with stutters when in use on my test system that clear when I don't run GPU-Z.

[GSalden 4,665]

44 minutes ago, Reset XPDR said:

New Test Version 0.4.2.7-test is available here. This test build aims to address stuttering issues that some have been experiencing since 0.4.2 was released, primarily by trialing the disabling of GPU-Z companion app integration as my own testing has revealed it seems to contribute to stutters. Please let me know if anything changes on the stutter front with this test build, particularly for those who have been experiencing them previously.

Changes are:

• Changed VFR mode to start up with TLOD Min like IFR mode does, rather than with the default MSFS TLOD setting like previously.
• Streamlined resetting of TLOD Min + on the ground when switching applicable settings or display modes.
• Changed TLOD Min + to use higher extra TLOD when TLOD Max is low
• Reduced TLOD Max in non-expert mode to 2 times default TLOD instead of 3 times, to reduce chance of stutters occuring at higher TLOD Max settings.
• Removed VFR mode TLOD sliding scale reduction as it can cause TLOD to be too high on landing and seems to cause worse stutters than if it marched down much quicker on that same approach.
• Disabled GPU-Z companion app integration temporarily, perhaps permanently as the information it gets is not used in the app for anything other than displaying GPU load, as it has been shown on a few occassions to correlate with stutters when in use on my test system that clear when I don't run GPU-Z.

Does these changes affect IFR mode too or only VFR mode ( only using IFR mode ) ?

[Reset XPDR 1,271]

2 hours ago, GSalden said:

Does these changes affect IFR mode too or only VFR mode ( only using IFR mode ) ?

IFR mode with expert settings is unchanged, except for the TLOD Min + change which only affect users who use a low TLOD Max but in a good way as the + was previously too small.

[GSalden 4,665]

32 minutes ago, Reset XPDR said:

IFR mode with expert settings is unchanged, except for the TLOD Min + change which only affect users who use a low TLOD Max but in a good way as the + was previously too small.

Here I use TLOD min + : 50 to be sure that I have max performance 

[Reset XPDR 1,271]

5 minutes ago, GSalden said:

Here I use TLOD min + : 50 to be sure that I have max performance 

If you always want TLOD Min 50, I suggest you don't check the + as it will add up to 50 more if your performance is really good (+15%).

[Vali 83]

8 hours ago, Cpt_Piett said:

So you get more stutters with AutoFPS off? Or do you just get mainthread frame time spikes that are visible in dev mode overlay, but no visible stutters?

I've spent a fair bit of time chasing stutters. Sometimes they're elusive and I can't for the life of me troubleshoot them. Other times, they're obvious e.g. during final approach into heavy scenery. Or even when overflying a small town at low altitude in a GA aircraft. What we have at the moment though, thanks to AutoFPS, is leaps and bounds better than what we had pre-AutoFPS!

I need to test it more but some conclusions on my side:

- shimmering is clearly due to VRAM, in some areas the 12GB is quite challenging (at 4K with DLSS Quality and DX12, no idea why MSFS Dev mode shows 10.6GB Memory). So here only 4090 or 24GB is the best option 🙂

- I reversed to DX11 (due to DX12 filling the VRAM sometimes), frame cap at 30fps (1/2 monitor refresh rate) and stutters seems start to occur on my side after the TLOD starts to decrease from the max set value (200) to the min value (50), also on the ground after landing (Main thread seems affected). Why, no idea... Perhaps I try to test it with fixed TLOD 50 all the flight and see if stutters occur.

Edited March 25 by Vali