The below shows the core usage on the default P3D flight without any add-ons at all as requested:
It doesn't show what I am saying at all - obviously no other cores are at 100% apart from Core 0 - but the core which is maxed out is running crazily hot and throttling - same scenario in 5.1, even with Core 0 at 100%, was running at 75 degrees Celsius.
So, is 5.2 working the CPU cores in a different way which is causing the extra heat and power draw is what I am trying to figure out.
Also oddly in the screenshot above, I had V-Sync, TB and TFC all disabled - no external limiters either which I would expect to be showing much more usage on all cores as the FPS wasn't being capped. Is it possible this issue has something to do with limiting FPS?
Thanks for your reply. I wasn't expecting LM to reply. I was somewhat frustrated so uninstalled v5.2 and reloaded v5.1.
However, given you're asking for info which could help solve things I will reinstall v5.2 but it won't be until tomorrow now. I appreciate the 8 hour time difference makes things difficult.
To answer your other question I run a BenQ PD3200U UHD monitor at 30Hz. VSync is enabled and frame rate limiter is set to unlimited. As I understand it the frame rates will be limited to the refresh rate of the monitor - 30 without any other software being involved.
More tomorrow when 5.2 is installed. I did try a quick test flight from EGLC with v5.1 and it was noticeable that the CPU1 load dropped into the 80s rather than 90+ and hitting 100 with v5.2.
Thank you for the reports. To help us better identify the cause of the issue, could you try disabling Adaptive SimGroups in your Prepar3D.cfg and post your findings regarding CPU usage?
When enabled, AI objects within distance and speed thresholds will simulate at the same rate as the user object. This is done to reduce jitter when in formation flight with AI objects or multiplayer players. Disabling this setting will leave AI objects at their predefined simulation rates.
Regards,
Clifton
Hi Clinton,
I'm attaching two screenshots showing CPU usage with TRUE and FALSE set in the parameter you mentioned. First with it set to TRUE, the default.
And with it set to FALSE...
I have a LUA script running on a separate PC using WideFS and monitor the number of Ai aircraft in P3D.
With the TRUE setting it stabilised at 79 on the ground. With the FALSE setting it was 173. I'm not sure why they should be so different but it's academic to some degree since the frame rates are too low to fly with such a high number of Ai in the sim. Hence why I use FSUIPC to limit them until 30fps is maintained. A minimum of 30 are retained in the sim and that's okay for all major airports.
I have a multi-core CPU and it's frustrating that so little use is made of the other cores. I suppose if it was easy to move the load onto the others you would have done it by now. I do have Hyper-Threading enabled as that was how the PC was delivered to me after being professionally built.
An update... Having just completed a circuit at EGCC I'm not happy with v5.2. The Carenado PC12 is not a demanding aircraft and AIG only had 35 aircraft at EGCC but with clear skies and on approach to 05L Core0 hit 100% on several occasions. Consequently fps dropped below 30.
That never happened with earlier versions. UK2000 is not demanding scenery so where is all this processing being done?
I'll stick with v5.2 because I have a dialogue with LM. But all is not well I'm afraid.
@Martyson, I’ll check tomorrow but I’m sure you’ll understand that a default aircraft at a default airport with only a smattering of Ai is hardly going to tax things.
As I said earlier the scenario I tested is not demanding of the CPU and with previous versions of P3D fluidity was always there especially when I’m limiting fps to 30.
My settings allow me to land a PMDG737 at fairly big airports with 60+ Ai previously. I’ll try other airports tomorrow.
Martyson wrote: ↑Tue Jun 15, 2021 3:51 pm
How is it with default P3D AI , default KRND (or other default airport) and default F-35A aircraft on one PC?
Martyson, I have a 6 core i7 and have set JobScheduler in P3D.cdg to 1365.
At KRND cold and dark in F35 with clear skies and 33 default Ai in the sim these are the readings using the same scenery settings as my test yesterday at EGCC.
GPU 21%
CPU 0 25-28%
RAM 11126Gb
Frame rates limited to 30 by using 30Hz on a UHD monitor with VSync enabled and Triple Buffering selected. Hardly a demanding test but provided as you asked for it.
If I slew up to 200ft and move forward at 20kts the numbers are:-
Completely fresh install of P3D v5.2 and my RTX 3080 is hovering around 5 or 6% usage according to Windows Performance Monitor whilst my i7-10700k is running around 90% utilisation.
I've just conducted a few tests on a flight from London City to Heathrow that overflies central London. EA on is awful for me. Stuttering mess and that's ignoring the visuals which are not to my liking.
I turned off all World / Scenery Objects. All sliders to the left. AIG Ai was injected but FSUIPC slowly deleted most until 30fps could be maintained. Only 35 aircraft at Heathrow. 100% core 0 hit multiple times. But with no scenery objects showing what is doing all the work?
Next was to keep scenery off and deactivate AIG Ai. Only default Ai running. 50 aircraft in a 80nm radius. Still multiple cpu 0 hits and relatively poor performance given my very modest settings.
Final test was zero scenery objects and no Ai at all, not even default. The result was a return to 5.1 performance. 100% core 0 hit only every 20-30 secs. 30fps maintained throughout. Fluidity!!
So it looks to me as though Ai is the culprit - even the default stuff. If anyone would like to try two tests - with and without default Ai and disable their 3rd party Ai that would be helpful.
