2 minutes ago, Martin EDLW said:

I have the same Intel.

Wich AM settings you use?

 

None, I switched to the other side now, MSFS. Sorry, I can not remember my settings.

There is something that I don't understand at all... and I've been reading a little bit in this forums. I've an AMD Ryzen 5 3600, with HT enabled in BIOS and no Affinity Mask set with Carenado 172sp and KORS orbx I've an overall of 30fps or less, and it get some stutters. If I set Affinity Mask to 62, per my understanding this is (HT[00]    Core[01]    HT[01]    Core[02]    HT[02]) enabled I've a considerable improvement to 60fps (I've v-sync and triple buffering enabled and my monitor is 60hz, so I assume maximum framerate).

Why this improvement? I've read in some sites that leaving the Core0 out of the sim gets an improvement, should be that the reason? should I test disabling HT in BIOS? but, will that impact in OS and other program performances, right?

I also want to test also AffinityMask=1365 to disable all HT cores leaving only the other 6 up, I'll update with tests

 

This is a party of try and failure 😅

20 minutes ago, ec793b said:

This is a party of try and failure 😅

It actually is nearly always because the systems are not the same, different hardware, different software, different addons ...

On my 18 core system (I need more than 16 cores to test software that uses more than 32 virtual cores) I forgot to add the affinity mask (AM) in P3D cfg the other week and P3D was a stutterfest. Without the AM and Hyperthreading on (HT) the first two cores do all the work as they emulate four cores that are shared between two physical cores and all the other cores only add to the load on the CPU.

For this machine I use AffinityMask=21845 (binary 01,01,01,01,01,01,01,01) which allocates 8 cores using the first virtual core from each and is a massive improvement to performance in P3D. With HT off I use AffinityMask=255 (binary 11111111) so as to only use 8 cores because adding more cores does not improve performance, only increases loads on the bus for no gain.

Hey Steve,
can you give me a value for my Intel Core i9-9900KS.
I would like to try this out.

Thank you

16 minutes ago, Martin EDLW said:

can you give me a value for my Intel Core i9-9900KS.

That's 8/16 cores I believe. So in P3D or FSX I would try 6 cores, leaving two cores free to the system:

With HT enabled, in the cfg:

[JOBSCHEDULER]
AffinityMask=1365

1365 decimal = 00,00,01,01,01,01,01,01 binary = 6 cores

In the binary representation with HT enabled, core zero is the ",01" on the right hand end, so as to use only the first virtual core of the first physical core.

7 cores = 5461

8 cores = 21845

 

With HT disabled:

[JOBSCHEDULER]
AffinityMask=63

63=00111111 = 6 cores

7 cores = 127

8 cores = 255

 

Regards
Steve

 

Does anyone here have one of the new Ryzen 5600X's and can give a recommendation for AM settings to try? It's a 6C/12T.

1 minute ago, Chapstick said:

Does anyone here have one of the new Ryzen 5600X's and can give a recommendation for AM settings to try? It's a 6C/12T.

With P3D I would use 6 cores so with HT enabled AffinityMask=1365, with HT disabled don't use an AM (or AffinityMask=0 or AffinityMask=63 each equate to six cores).

15 minutes ago, SteveW said:

With P3D I would use 6 cores so with HT enabled AffinityMask=1365, with HT disabled don't use an AM (or AffinityMask=0 or AffinityMask=63 each equate to six cores).

I'll try that 1365 as my processor is 6C/12T also, I'll check results

Quickest way to calculate AffinityMask to disregard HT and only use real cores is to total count threads (9900k/10700k) 16 threads then divide by 4. That gives you 5555 hex value then convert to decimal which is 21854. (8700k Ryzen 5600 etc) would be 12 threads giving 555 hex which is 1365 and so on.

Of course this doesn't work when using combo real and HT threads. Also easier to remember than having to go from right to left in which earlier posts i made caught me posting wrong.

Edited January 8, 20215 yr by 40track

On 1/8/2021 at 8:16 AM, SteveW said:

With P3D I would use 6 cores so with HT enabled AffinityMask=1365, with HT disabled don't use an AM (or AffinityMask=0 or AffinityMask=63 each equate to six cores).

This seems to help with stuttering issues I've been having a lot with the 5600x. Unfortunately, it causes a hard pause every minute or so in the FSL 320 only for some bizarre reason.

It's quite hard.

What's the best setting for P3Dv5? I've an 10900K (at 5.1 GHZ ) with HT off at the moment. Some stuttering.

Has anyone it running with HT on and AM? Which AM setting do you have?

Thanks for your help

For those using Ryzen, I’ve tested this quite a bit on my 3700X and in V5, I found no affinity mask setting in the P3D config file with SMT Off works best, at least for me. And for those using FSLabs, that will set a job scheduler in its config file anyway, assuming you tick tha option, which I do use

1 hour ago, Urmel81 said:

It's quite hard.

What's the best setting for P3Dv5? I've an 10900K (at 5.1 GHZ ) with HT off at the moment. Some stuttering.

Has anyone it running with HT on and AM? Which AM setting do you have?

Thanks for your help

Don't have that CPU but copy/paste to Prepar3d.cfg

[JOBSCHEDULER]
AffinityMask=349525

With your HT on 

 

Edited January 10, 20215 yr by 40track

4 hours ago, mrlip said:

For those using Ryzen, I’ve tested this quite a bit on my 3700X and in V5, I found no affinity mask setting in the P3D config file with SMT Off works best, at least for me. And for those using FSLabs, that will set a job scheduler in its config file anyway, assuming you tick tha option, which I do use

Yep, looks like you're right. SMT off, no AF mask in P3D cfg and now I seem to be getting fewer stutters than with SMT on and FSL is behaving better (with their scheduler config files).