P3DV5.3 vs P3Dv5.4 what is the difference?

[Ray Proudfoot]

The SDK for v5 is here. I have reassigned the MainThreadScheduler to core6 from 0 to keep the latter free for OS duties.

The fix for aircraft wheels at FSDT airports is not one that bothers me. I couldn’t see any other benefits I lost as a result of reverting to 5.3HF2.

 

[Sparkrite]

46 minutes ago, 787flyer said:

If i am not mistaken, in v5.3 you could assign the P3D main thread to any core (whatever is your fastest core for your CPU) but in P3Dv5.4 I thought it was locked to core 0 (so I read), but I stand to be corrected as I never did install v5.4

If that's not the case, that would be one less obstacle in moving to v5.4.

Yes indeed, I suspect you are mistaken @787flyer. The main thread can be assigned to whatever core you like in V5.4.

Here is my AF for V5.4:-

AffinityMask=4294967295
P3DCoreAffinityMask=4294967292
MainThreadScheduler=15
RenderThreadScheduler=11
FrameWorkerThreadScheduler=28

 

[Luke]

1 hour ago, 787flyer said:

No, I dont think so Luke. I have a Job Scheduler that can assign the P3D main thread, the terrain (Renderthreadscheduler) threads and Frameworkerthreadscheduler to different cores in P3Dv5.3

You're right - for some reason I thought that came out in 5.4.

But Sparkrite is correct - the main thread can be on any core you want - in fact an application cannot force a thread to a specific core in Windows. It can always be overridden.

Cheers

[787flyer]

Thanks Guys for the feedback on the Job Scheduler uncertainties. Looks promising.

So,  I did some more digging into the EHAM textures issue in v5.4 and this can be overcome by updating the LUA script.

I think the only other thing that prevents me from moving to v5.4 is the issue Ray ran into with AI traffic causing some spurious issues, if I remember.

 

@Ray Proudfoot- did you ever get to the bottom of this?

[Ray Proudfoot]

44 minutes ago, 787flyer said:

@Ray Proudfoot- did you ever get to the bottom of this?

Nope. Very difficult to determine the BGLs responsible.