Thank you for all of your continued input.
It’s obviously an msfs problem, despite the poor support I received through the support function.
Please keep sharing your thoughts and hopefully asobo look at this seriously.
1 Like
I too am having the exact same problem. It does not appear on the ground and only appears during the ascend-descent cruise. I’m always nervous about CPU usage and temperature while flying, and it only appears when I’m flying in Europe. This phenomenon has appeared since around SU14. Before, I was dissatisfied with the CPU not working, but now it gets too hot. I’m worried. I’m still flying over Italy and the CPU temperature is over 80 degrees. This is clearly a problem in Asobo. It didn’t happen before and there is no problem with other games.
Yes, and the silence with which these reports are received is impressive.
When I set the rolling cache to 32GB, the problem became a little less serious. It still rises occasionally, but it returns to normal quickly. It’s not a solution to the problem, but I hope it helps.
Interesting. I have mine off as I have highspeed internet. I wouldn’t love to turn it on but may test… Could you report back?
I also use 1G high-speed internet. However, the CPU was suffering, and when I used rolling cache, it seemed to be fine for a while, but after a few flights, the overheating phenomenon reappeared.
In other words, rolling cash doesn’t seem to be a separate option.
After that, I started looking for other causes.
Meanwhile, there was a file called Cost line Fix…? (I forgot the exact file name because I deleted it) in the community, so I deleted it and the CPU overload disappeared.
There have been no problems on my three European flights so far. I’ll have to wait and see a little more, but I installed and forgot about the file that fixes the European coastline a long time ago (early 2020). I thought this file might be causing the problem, so I deleted it and it’s fine.
If you’re not sure, check to see if this file is in the community folder.
After deleting this file, I run 2020, deleted the rolling cache, deleted the scenery index file, then run 2020 again and quit. And after the flight everything was fine.
I come a cross this thread as I was thinking there is something wrong with my build, I recently upgraded to 7800x3d with 4080 super, when flying fenix or PMDG planes in Europe is where I mainly fly getting this spikes in temps. Absolutely crazy if you ask me, running benchmark for well over hour on all cores highest temp I got was 81c, run MSFS at cruise altitude 60-65 then every so often you get 87c. I have to set fan curve to higher just so I don’t get the fans spinning up and down, Currently flying in US temps hovering 60-66 but crucially no spike in temp every few seconds. Anyone found solution?
Currently, there is an issue flying over Europe in MSFS.
The problem manifests as stuttering and performance issues, which particularly affect CPUs including X3D models.
I recommend limiting the FPS in the NVIDIA Control Panel and the MSFS profile. As for your CPU if you’ve switched from Intel to AMD you might be surprised by the Ryzen 7 8700X3D behavior. It’s known for having temperature spikes that last for a second or two. This is normal so don’t worry as long as it doesn’t exceed 90 degrees Celsius.
edit
My Ryzen 7800X3D behaves in exactly the same way, with the difference being that during flights over Europe the temperature might spike to 83/85 degrees Celsius for a moment.
I’m on air cooling and it does it for a few seconds periodically. I have my fan curve very being very linear up to the low 50’s then spiking aggressive in the high 50C range(around 85%) and going to 100% at >68C to catch it and it calms right back down. Most of the time the sim has the CPU in the mid 50’s and the fan’s barely working. It’s an MSFS thing. I can’t find any corresponding ‘reason’ for it. Got sick of watching it frankly. Can’t make heads or tails if it has anything to do with the occasional freeze-up(which seems to have no correlation to what’s being displayed either). I have flights that have two freezes and spikes and 4 flights with nothing. It can happen in a Cub in the middle of nowhere and not have any issues in a big city with a full glass complex bird in much heavier traffic and ships. I’m so ready to move on to 2024.
1 Like
Been honest don’t get them spikes anywhere else, I am aware of x3d chip having small die which heats up rapidly that’s not the problem as AIO will keep up but let’s say just for example you get temp spike every 10-15 sec and highest I have seen was 88c that’s borderline of throttling in that cpu, I am running well undevolted is well with each core PBO set individually, it doesn’t have problems reaching max core speeds, don’t get this behavior in other sim like xplane, haven’t seen this behavior flying in US last night in MSFS. Will limit the frames to see if it has any effect.
Otherwise I have no problems with temps as a rule sim runs around 60c with 360AIO and no spikes.
Thanks for your reply and advise.
Have you experienced this everwhere or just Europe?
Thanks for posting. The devs continue to be very quiet about this. I can assure you it’s not your hardware.
In answer to your second question, mostly Europe yes.
I wouldn’t worry about short temp spikes like that, even if they hit CPU throttling limit. Even AMD states that this is to be expected.
This is simply how the CPU reacts to this particular usage scenario. In case of MSFS, I noticed that it can heat up the X3D processor in short bursts, especially when loading lots of photogrammetry scenery with custom POI.
Modern CPUs have pretty complex power management which dynamically adjusts the voltages and clocks based on CPU load. If CPU is not fully loaded, the voltage will rise in order to boost the clocks on the few cores that are doing some work, as long as we stay within the total power and temperature budget.
I don’t know what exactly MSFS does, but it seems that it produces CPU load that is low enough to increase the CPU voltage, but high enough to considerably heat up some parts of the CPU, which causes those temp spikes.
Don’t think MSFS devs can do anything about it, other than intentionally slowing the scenery loading and other MSFS activities to reduce CPU load. But that would be counter productive. It’s up to CPU manufacturers to ensure that they can handle everything the PC software can throw at the CPU, and throttle back accordingly if needed.
You can increase thermal mass on the CPU so the energy has a place to go without raising CPU temperature as much. One way to do that is with water cooling and the water-filled heat sink/pump which also has a fairly large reservoir/radiator to spread the heat out over more mass which means lower temperature. Particularly large/heavy heat sinks with high thermal conductivity (eg copper - not aluminum) with fan(s) can also help. High thermal conductivity thermal paste properly applied is also important.
CPUs can go from idle to full speed really fast. Fans can’t ramp up fast enough to remove the heat. In the time scale the CPU can ramp up, the fan barely moves. The load comes up in micro to nanoseconds. Some may end up having to run their CPU fan(s) at full speed when running the sim to lower the CPU temperature and give a little more headroom for spikes plus the fan doesn’t have to speed up for maximum cooling - it’s already there.
Wait until FS2024 hits. With more cores running hard, average CPU temperatures will increase and cut into margins for handling spikes without overheating and throttling. I think some may end up being surprised with the extra power demand and heat that’s coming.
One thing to be on the lookout for, though - do the cores automatically “un-throttle”? Some systems at least used to take a reboot to clear throttling and get the CPU back to full speed. It’s been a few years since I was involved in this issue but we had to reboot systems that throttled to get them running full speed again.
I’d hope that has changed but it might not have. Reboots to clear are a pain in the tuckus.
This won’t help as the problem is the heat transfer from transistors layer to the CPU surface, not the capability of the cooler on top. These CPUs can easily reach thermal limits even under the best water cooling, even though they draw less power than the older generations and stay relatively cool on the outside.
No, it’s nothing like that, the CPU will simply drop the frequency and voltage slightly to bring the temps below limits and resume normal operation as soon as there’s some thermal headroom. Modern CPUs (and GPUs) adjust their working parameters hundreds of times per second, temperature is just one of the inputs to the algorithm.
2 Likes
As mentioned earlier get on those fan curves, in the bios, by a program but get ahead of the heat.
I have an even 3-in 3-out 2 on the CPU for fans so I have them grouped as such but they may as wall be on the same header, 40c 40% - 50c 60% - 60c 80% - 70c ROAR 100% but it’s summer and I have AC and fans louder than my PC but even in silence at 100% it’s not much louder than an old woman aggressively blowing on a hot bowl of soup. 
Same thing with your GPU, get the fans spinning early, any amount of flight sim noise or even a headset should block it out anyway.
Well I tough found a fix by removing world update XIII but flying from Dublin to Gatwick still does it, Will have to put up up with this for now.
This was a very articulate description, thank you for that. I think that will put a lot of people’s mind at ease.
For me the CPU is, as you say, behaving as it’s designed to, we’re not damaging the CPU. However, i do hope there is a software solution for this, as I genuinely do not think this should be happening like it is.
I was setting my per-core undervolts using CoreCycler, which stresses each core one at a time. I used a hardware monitor to guide my increases in negative offsets, looking for the point where the max frequency during the test matched the max boost frequency spec.
Imagine my surprise when I saw one of the cores had a max temp of 203°C during one of the core cycles. No apparent damage, and everything’s running really well now, with <60°C temps during normal flights. I’m guessing that must have been a sensor glitch.
Could a CPU survive an actual spike to 203°C? I honestly don’t know.