Another P3D4 and HT (well, SMT) question

[Luis Hernandez 581]

Dear all, I've been reading all  @SteveW´s posts regarding Affinity Mask in P3D v4.5. I've finally understood that, while HT off + no AM would be easier to setup, HT on with an adequate AM will yield better results. Also, I understand I need to send my addons (namely PF3 and ActiveSky) to the last core.

For my laptop (i5-8300H, 4 cores, 8 threads), I have no option but leaving AM on. According to this post from SteveW, AM=85 (for reproducing HT off) is a good value, and AM=245 (leave Core0 and Core1 with only 1 LP for P3D) is even better. Now, I've just built a desktop with a Ryzen 5 5600X (manually overclocked to 4.5 GHz), 16 GB RAM and a GTX1650 (not paying more for a GPU these days), using an aftermarket CPU air cooler. I tried without any AM, with both SMT on and off: with SMT on, I get almost 7-9ºC higher, but I felt a smoother sim (very important when approaching KJFK 31L with 100% AI and real weather). However, I'm not sure how to extrapolate the 4-core AM=245 to a 6-core CPU:

- I can just fill the additional cores with 1s, i.e., AM=11,11,11,11,01,01=4085

- Somewhere in your posts I read that P3D can dedicate up to 3 cores for itself, and then spread and share workload for all other cores. I.e., AM=11,11,11,01,01,01=4053

- Just assign 1 LP per core to P3D, similar to AM=85 in a quad-core. I.e, AM=01,01,01,01,01,01=1365.

 

I keep FPS limited at 30 via RTSS, and unlocked in the sim. Since each system would be different, how should I test to get the best AM for me? I mean, I know the input variable will be the AM, but what should I be measuring as an output? I could test CPU temperature peak or just do a subjective evaluation of P3D's smoothness and loading times. What do you suggest? I'm lost.