Mainthread loads up more and more and doesn't recover

[PaRaBaS6435]

First some Info’s to my hardware:

R7800X3D watercooled 50 degree celsius while MSFS24 is running, 64 GB Ram 6000Mhz Cl30, RX7900XTX Sim on a SSD.

I’ve observed it several times now. Whenever I fly from one airport to another, everything is fine at the dep airport: main thread, RDR thread, GPU, everything is in the green (dev mode).

I have to say that it doesn’t matter where I’m flying from or where I’m going to; it’s always the same.

Today, for example, like the last few times, I took off from Frankfurt (Aerosoft) to Hamburg (Simwings), and everything was great. The longer I flew, the higher the main thread timings rose. Upon arrival in Hamburg, the timings were significantly higher than in Frankfurt during takeoff. I continued the return flight with the Fenix A320

​​without reloading, and once back in the air, the main thread continued to rise. When I arrived in Frankfurt, it all came to an end. Main thread timings of over 25ms upon landing.

I took off with main thread timings of 10-11ms in Frankfurt. The longer I was in the sim, the worse it got.

Shortly after takeoff in EDDF:

Arrived in EDDH

A screenshot from the landing back in EDDF I have not, because everyzhing was freezing.

But I could entcounter, that the mainthread doesn’t recover. It loads up and up and up and that made the worst thing happen back in Frankfurt.

So something is defenetly wrong. At least in the air, the timings should recover, but stayed fully loaded.

Timings back in EDDF near 30ms, another 10ms on top of the 20ms in Hamburg.

Only addons are airports and the A320 family by fenix as well as INI A350 + GSX.

[CasualOne7471]

What version of the airport? I had the same thing happen at a 3rd party vancouver scenery. I deleted the scenery and everything has been fine since.

[PaRaBaS6435]

Frankfurt by aerosoft latest official 24 supported version and simwings hamburg.

I did not entcounter that in SU2.

[RefluentBrute7]

I believe this is not an isolated problem. I have been running some tests in this regard and the impression I have is that the mainthread is unable to recover at some point. I use PSXT and when I increase the radius to 100 mn, the amount of traffic increases a lot and when I close the application, the mainthread does not recover to the same previous levels even with a lower load. I suspect that it is not a problem related to the scenery. I also thought it could be something related to Fenix, I reinstalled it but the problem persists.

[Markucha21]

It’s not isolated problem it is another very very big issue form ms/Asobo repair one destroy another they never will be able to get out from this beta simulator 2024. I have the same issue after minutes cos drops from 90 to 50! Without reason and the mainthread limited is huge

[PaRaBaS6435]

Yep, I also do think it isn’t related to the scenery. I think it is a deeper issue inside the core. That could also be the issue some are facing, (stutter on landing) because mainthread gets hammered and is unable to recover what should give us the headroom for landing at airports.

I’m back to SU2 right now, tested a lot these days. I hope this will be fixed before SU3 drops, otherwise it will be loud in here.

[RefluentBrute7]

I must have gone back and forth to SU2 at least 10 times. With Build .13 it became a little more stable but I noticed this issue that you mentioned in the mainthread. If I remember correctly, this also happens in SU2. I think it would be important for this issue to be voted on a lot to move up in the list of issues.

[PaRaBaS6435]

Over the weekend, I will test SU2 vs SU3 b.13 and compare the numbers in the exact same scenario.

I will post the screenshots here. But you are right, it needs more votes to get more attention.

[PaRaBaS6435]

I couldn’t wait, so I did a second test with SU2 stable.

Same flight from Frankfurt to Hamburg. Same aircraft Fenix A320.

This time in SU2, no mainthread issues.

Again shortly after takeoff mainthread is even lower than in SU3 b.13. See first post!

You can clearly see the mainthread after landing in Hamburg is much much lower than in SU3 b.13. Also it didn’t rose in the air.

Now back in Frankfurt.

Back in Frankfurt and what can I say. Much much better, no system freeze no stutter.

I dont know if the mainthread issue in SU3 b.13 has something to do with the new LOD curve or what ever, but SU2 runs a lot better on my system.

Btw: Frames are locked because of FG.

[RefluentBrute7]

Same conditions and completely different results. If there was any doubt that there was a problem with the main thread, it’s now cleared up. I think the point is very relevant, and a simple test is valid. Load free flight anywhere and wait 10 to 15 minutes, and you’ll notice the increase in the main thread and the decrease in FPS, even if you don’t load anything on the SIM at all. In my tests with the Fenix ​​in any condition, I always start with FPS around 80, which after loading stabilizes at 90 and then reaches 68-70 FPS. I think it would be important for more people to do this test so we can confirm.

[NixonRedgrave]

To confirm the repro steps: simply load in and sit parked while time passes?

[RefluentBrute7]

In my case, I only needed these steps to confirm the regression. I’ll download build .15 now and see if I can confirm it and take screenshots so we can discuss this. I’ll also test without the Fenix ​​to see if it’s related to the aircraft.

[rick12251]

I get this same issue in SU3 beta with live traffic on. mainthread increases by 10ms or more as time goes by. Turning off traffic instantly returns mainthread to normal.

[NixonRedgrave]

@RefluentBrute7 thanks for the input.

@rick12251 thanks for that further detail. I’d noted with .13 a 50% decrease in FPS with real-time-online traffic and ground traffic on and set at Ultra (granted I have the FSLTL models installed). I wonder if the decrease in FPS is due to this mainthread issue. I’d been sitting at KSMO (auto-gen) for a spell when checking this out.

I will do some investigating on this issue myself, shortly.

[RefluentBrute7]

Could the issue be traffic-related after all? I’ll run a specific test on this under the same conditions on build .15.

[BragRaindrop933]

It sounds a bit like what I saw when I just sat at the gate and restarted a flight. I can boot in to any flight and start it & I get the best performance (that my system can delivery anyway). If I do nothing (not even turning aircraft battery on), but go to pause menu and restart flight, I lose a whacking amount of mainthread performance (& FPS), which doesn’t ever recover. If I restart flight again and i lose some more. The only way I found to recover performance is to reboot the sim. There is definitely a performance (mainthread) degradation issue in FS2024, and I’m wondering if this is what you are seeing albeit in a different manner.

[PaRaBaS6435]

Yeah what ever it is, apperently it is also present in update .15 as I saw a post in PC - Performance survey. Someone else mentioned heavy mainthread issues.

In my case, I’m using vatsim for my test but limited the amount of aircraft to 30, that my system isn’t overloaded. Maybe FSLTL aircraft models?

But I haven’t that problem in SU2, as you can see.

[BragRaindrop933]

Not unexpectedly I noticed The Fenix is a lot higher mainthread wise that either the default Max or the pmdg777.
I was using Beyond ATC injecting FSLTL models the other night with the Fenix at EGLL (Gaya) and mainthread was getting hammered. I was seeing FPS of 13.
Can’t help but feel that performance has regressed in the sim since release.
FS2024 was meant to load multi cored CPUs better and free up mainthread. In SU1 I was able to fly without external FG.(I have a 3090). But now with SU3 I’m having to shoot for a target of 20 FPS and use external FG (lossless scaling) to get 60. Which was exactly what I was doing in fs2020, so so much for those touted fs2024 multi CPU core mainthread optimisations!

I digress though, as there is definitely is a performance degradation issue in the sim. The performance issues pains me as it’s really restricting my enjoyment with airliners right now, which is mainly what I fly, and I refuse to go down the new hardware path as it was performing better and totally usable in SU1.

[RefluentBrute7]

I ran some more tests with build .15. There was no degradation (if we can call it that) with the Fenix ​​idle and turned off for a 15-minute interval. I tested with PSXT to inject traffic, with 29 aircraft on the ground and 16 active. I let it run for a few minutes and then closed the injector. There was no degradation. I ran a new test with PSXT on and rolled the runway (I’m testing in KMSP) across the entire runway and aborted at V1, returning to the gate. In this situation, I observed not only FPS degradation but also an increase in the main thread from 20 ms to almost 30 ms. The amount of traffic remained stable the entire time. Under these conditions, it seems to be something truly in the core, not related to the menu or restarting the traffic or aircraft.

EDIT: We need more votes to get attention for this topic. It’s a super relevant point and deserves the devs’ attention. This is something that completely kills immersion and makes it simply unplayable.

[NixonRedgrave]

Interesting.

I’ve had the PMDG 772 powered on at the gate at default hand-crafted KLAX for a long time.

I ran without traffic and saw zero change in FPS.

I’ve since turned on ground and realtime air traffic at ultra with FSLTL models and while FPS dropped 20-40 FPS depending on which view I’m in, it has held steady and consistent.

I haven’t tried your aborted takeoff procedure to test for degradation.

Mainthread has an occasional red blip, but otherwise nothing massively alarming.

4K native, TAA, FSR 3 FG, Dynamic On, Ultra.