Doesn't it make sense to let P3D have all cores and take all the other programs off of core 1, where the thread manager lives?

5 hours ago, denali said:

Doesn't it make sense to let P3D have all cores and take all the other programs off of core 1, where the thread manager lives?

The thread manager does not live on core 1 or core zero. Also there will come a time when there are just too many cores even without enabling HT, so they would be limited to requirements in many cored CPUs.

On 10/22/2017 at 7:31 AM, SteveW said:

Thing about four cores is that you need them all for P3D v4 so use the AM to disable just the first LP with HT enabled or with HT disabled do not use an AM and the sim uses all four cores.

Hi Steve,,

So with a quad-core processor, w/ hyperthreading ON, one would use a AM  = 254? Is there functionally any difference between AM 253 vs 254?

Mike

 

 

2 minutes ago, orbmis said:

Hi Steve,,

So with a quad-core processor, w/ hyperthreading ON, one would use a AM  = 254? Is there functionally any difference between AM 253 vs 254?

Mike

 

 

Yes we might use such an AM. Those two AMs amount to exactly the same thing to the sim. Core zero has one Logical Processor enabled for the sim. To the sim there are seven LPs available which are loaded by the jobscheduler as the sim makes each process depending on the split-out. Just looks like a seven core CPU irrespective of LP count assuming there are cores or LPs enough to go round.

3 hours ago, SteveW said:

Yes we might use such an AM. Those two AMs amount to exactly the same thing to the sim. Core zero has one Logical Processor enabled for the sim. To the sim there are seven LPs available which are loaded by the jobscheduler as the sim makes each process depending on the split-out. Just looks like a seven core CPU irrespective of LP count assuming there are cores or LPs enough to go round.

Can't resist jumping in here... I've been using either AM 252 or 254 with a .bat file that starts my modest 3 add ons (FSXFlight, PilotEdge client & ASP4).  With the AM 252 [11, 11, 11, 00] I use a .bat with affinity of "3" [00, 00, 00, 11] and with AM 254 [11, 11, 11, 10] I use a bat affinity of "1" [00, 00, 00, 01].  I've test them both (and many other combos) logging CPU & GPU utilizations and temps and monitoring FPS.  The results haven't been overwhelmingly conclusive but always better than no AM and no .bat affinity.  My add ons seem to use very little CPU time so the 254 / 1 combo seems best to me.  Any reason why that shouldn't be good?  My 4790K specs below with HT on.

58 minutes ago, bbuckley said:

Can't resist jumping in here... I've been using either AM 252 or 254 with a .bat file that starts my modest 3 add ons (FSXFlight, PilotEdge client & ASP4).  With the AM 252 [11, 11, 11, 00] I use a .bat with affinity of "3" [00, 00, 00, 11] and with AM 254 [11, 11, 11, 10] I use a bat affinity of "1" [00, 00, 00, 01].  I've test them both (and many other combos) logging CPU & GPU utilizations and temps and monitoring FPS.  The results haven't been overwhelmingly conclusive but always better than no AM and no .bat affinity.  My add ons seem to use very little CPU time so the 254 / 1 combo seems best to me.  Any reason why that shouldn't be good?  My 4790K specs below with HT on.

Seems fine.

With P3dV4 having improved the split-out over earlier versions shows the second process a very small and so affecting the use of the first sim core less than with older sim variants, giving good results with 252 on three cores even when compared to 254 giving four cores and the one LP of the first sim core.

The less change that can be noted between them, the more likely something else is flat-lined. Also beware that activity created by the resources implemented by the addons or P3D don't necessarily run on the cores specified in the AMs, they will fill any unused hole. It is generally this activity moving around gives the appearance of changes to the performance.

But as you noted - although the carefully constructed AMs do yield a better overall result than leaving to chance, there's little difference noted between three cores one shared on the main thread verses four cores no sharing the core of the main thread.

So giving that noted above as an example, how can an 8 core CPU help? The performance noted seems little change if any between three cores or four in use by the sim, so why use five, or six?

Using too many cores consumes the resources far too readily for no actual gain in performance. Therefor it might be that if we have an eight core CPU we would run into performance issues if we avoid the use of an AM, HT On or Off. Use the excess cores to locate addon activity.

Steve your my goto man for AM cores etc.

I don’t use a AM anymore because you said no need in v4.

But I do kick ASv4/ACSA/chaseplane and everything else of core 0 and 1

Is that a good idea?

I am 4 cores with HT ON

Yes should work fine, seems like you got it covered. Check that if the first LP hits 100% you can reduce the fps limit, reduce the settings by 17%, or use an AM with '01' on the right end to reduce the use of the first sim core by an unnecessary 17% since with '11' on the end the core is shared with the second process.

Thanks Steve,

I saw my core 1 was at or close to 100% a lot so that’s way I kicked everything else of it.

That's right. That's the main thing, keeping stuff away from the core running the main job of updating the screen. The other processes in the sim can take several seconds to complete and so the sim will be less affected when they are held back slightly with addon activity.

3 hours ago, Nyxx said:

Steve your my goto man for AM cores etc

+1  Thanks Steve, always insightful.  

Just for clarification: the add-ons should be moved away from the first core, which is core0? Sometimes it is confusion when people state core 1 and it is not clear whether they mean the physical core0 or core1.

I use PL. So for a four core CPU is it advisable to move the addons to a single core (core1, core2 OR core3), or better to move them to several physical cores such as core2 AND core3?

And what is the reason behind this: I used to use AM=85 since FSX days with my 3770K HT on. Always got the best results regarding blurries, FPS and smoothness. Same is true now for P3Dv4.1, however, the difference to no AM at all is barely notable. And I never did care at all about my other software, never used PL or bat files. Yet I think my FPS and performance is perfectly fine...

Does Process Lasso change AM on the fly?  Can't you just play around with that to see what works best while flying around?