11 minutes ago, Dirk98 said:

When not using HT just remove the whole section [JOBSCHEDULER]. Prepar3D.exe runs very well on all 8 cores of your CPU and loads a little faster than on 6 cores.

thanks.  I don't have it now.  Just was wondering if it was something I could benefit from adding.  

very strange, uninstalled the client again, delted p3d.cfg again reconfigured everything frames are back to the same level as before.

 

maybee it was one of the settings from the rd presets p3d 5.2 guide.  Didn`t apply any tweaks to .cfg till now, and everything works fine so far.

Edited June 22, 20214 yr by 331BK

17 hours ago, Luke said:

Fair enough. So I don't put words in your mouth, how do you believe threads are created and scheduled in FSX/P3D?

Cheers

The discussion became one of setting affinity because Ray's config didn't produce the correct result. I noticed that there was a problem in his config file so I joined the discussion and the typo error in the config was corrected. Ray needed to correct that before continuing his evaluation of the problem he observed (outlined in the discussion topic heading).

The reason I came back to this discussion about P3D affinity is because I was asked by Ray if it was useful to use the affinity mask config setting with an HT disabled CPU, which I answered. But otherwise the discussion has veered off topic. (Your question is not a part of this topic and I see no reason to enter into an off-topic discussion about it just to satisfy your demand.)

During this time I, and others, have only been talking about what we observe when we run P3D (and FSX), nothing to do with how it is done, which is irrelevant.

We observe that after running P3D we see that same activity across the used cores (or LPS). At no stage did it matter how that was done by the system of Windows, the jobscheduler and the simulator program.

It is obvious that the P3D program starts, reads its config settings, determines an affinity if it is included in the config, and sets affinity from there. We can go to Task Manager, Details, Set Affinity, and there we can see which cores (or LPs) have been selected for Prepar3D. If we set the affinity in the config we see the result there with Prepar3D.exe. Looking at Task Manager performance graphs we also see the activity in the cores we chose in the affinity setting.

The difference with FSX is that even though we set affinity in the config, and we can see that in the Task Manager graphs it is done as we desire, we see in Task Manager, Details, Set Affinity, that all the checkboxes are still checked. Even so, only the cores we unmasked in the affinity setting are showing the activity. So the activity per core, or the "Task" per core (as I put it for simplicity of description)  has remained fixed, or pinned to the starting core we desired.

What we observe is the same activity in the unmasked cores (or LPs) no matter how many times we start the simulator and no matter what else is running in the system.

We notice that if the first core of activity, the first Task, is encumbered by another process, the performance drops, or fps reduces. We can assume that the first Task is more concerned with fps. We notice that if we encumber other cores of P3D (or FSX) the background process performance drops, the fill-in rate around the player reduces.

So we have mentioned that the best thing to do with regard to maintaining performance is keeping other processes away from the first task a priority, and others in general.

there seems to be much discussion about Affinity Mask settings - for P3D 5.2 - when many are happy with 5.2 performance now without such AM tweaking,

only once was I able to get 100% core0 usage when Super Traffic Board was used - have used it subsequently with no problem…

Process Lasso is used for CPU management & works very well,

no need for any AM tweaking here.

Edited June 23, 20214 yr by craigeaglefire

19 minutes ago, craigeaglefire said:

Process Lasso is used for CPU management & works very well,

no need for any AM tweaking here.

Since Process Lasso is used to assert affinity settings, you are really saying you are tweaking AMs, but maybe not in P3D. If you scroll up a bit you can see that no AM has been recommended for P3D in one system, but then in Ray's system, and in some other contributors systems, they find that an appropriate AM helps P3D run better.

So in fairness, perhaps it is better to say that it is system dependent whether an AM for P3D is helpful or not, and it could be that in some systems, AMs are helpful when asserted on other programs, and that Process Lasso can be used effectively for that purpose.

14 minutes ago, craigeaglefire said:

there seems to be much discussion about Affinity Mask settings - for P3D 5.2 - when many are happy with 5.2 performance now without such AM tweaking,

only once was I able to get 100% core0 usage when Super Traffic Board was used - have used it subsequently with no problem…

Process Lasso is used for CPU management & works very well,

no need for any AM tweaking here.

@craigeaglefire While, of course, we are happy for you, it would be more helpful if you could bring yourself to spare a thought for those of us who, despite our years of hard-won experience, are less fortunate and have found solutions by employing the affinitymasking capabilities of this sim.

However, understanding the nuances of the affinitymask, as applied to modern processors in differing hardware/software configurations, is vital for a successful outcome. In this we owe Steve a huge debt of gratitude. I’ve lost count of the number of approaches he has employed in this forum as he continues his attempts to educate us. His patience is extraordinary!

One thing is clear and that is that this can be a difficult concept to get your head around. Happily, once understood, there is little doubt that an intelligent use of the affinitymask can reap many rewards in terms of improved performance and reduction/elimination of unwanted stuttering.

Regards,

Mike

With an AM setting of 1365 this is a surprising use of my 12 VPs.

@SteveW, I'm flying off the coast of Morocco from Casablanca to Malaga. Very little traffic around. Aircraft is XTreme Prototypes Lear 25.

Why would so much processing be passed to cores 2 6 and 10? My other programs are using 9 and 11. Core 0 far from being fully utilised. I'm not complaining because everything is ultra-smooth. Just curious.

Flying over water .... not much to render.

I'm not saying that this is what is happening to you but blocking P3d from using every other virtual core does not prevent either the OS or other active processes from using those "blocked" virtual cores.

11 minutes ago, Ray Proudfoot said:

Why would so much processing be passed to cores 2 6 and 10? 

I think you will find those LPs are utilised by scenery collecting. I get exactly the same graphs as your screenshot with AM1365. So I did a quick comparison with minimal scenery settings, and I get far less use of 2, 6,and 10. Our systems are very different but P3D v5.2 acts the same.

11 minutes ago, jabloomf1230 said:

I'm not saying that this is what is happening to you but blocking P3d from using every other virtual core does not prevent either the OS or other active processes from using those "blocked" virtual cores.

Yes that's correct. Since the P3D AM can only mask LPs from P3D using them, that is why we tend to find better results by corralling other exe apps by asserting an AM on them, such as addons running alongside the sim, to direct their use of the CPU onto LPs away from the main LPs.

1 hour ago, Ray Proudfoot said:

With an AM setting of 1365

My other programs are using 9 and 11.

Ray,

You could put your other programs on LP5 and LP9 freeing up LP11 which is taking throughput away from LP10, and only adding to the loads on cores 2 and 4 where LP4 and LP8 are hardly used.

Edited June 23, 20214 yr by SteveW

54 minutes ago, SteveW said:

Ray,

You could put your other programs on LP5 and LP9 freeing up LP11 which is taking throughput away from LP10, and only adding to the loads on cores 3 and 5 where LP4 and LP8 are hardly used.

Thanks Steve, I’ll try that tomorrow.

1 hour ago, Ray Proudfoot said:

With an AM setting of 1365 this is a surprising use of my 12 VPs.

@SteveW, I'm flying off the coast of Morocco from Casablanca to Malaga. Very little traffic around. Aircraft is XTreme Prototypes Lear 25.

Why would so much processing be passed to cores 2 6 and 10? My other programs are using 9 and 11. Core 0 far from being fully utilised. I'm not complaining because everything is ultra-smooth. Just curious.

With this distribution of load across LPs I would consider your initial problem fixed, Ray. Looks very similar to what I'm seeing in v5.1 currently. So was it just change of AM that did it?

Edited June 23, 20214 yr by Dirk98

2 minutes ago, Dirk98 said:

With this distribution of load across LPs I would consider your initial problem fixed, Ray. Looks very similar to what I'm seeing in v5.1 currently. So was it just change of AM that did it?

I changed the title to include [Solved] but some felt that because they still had issues that was misleading. I removed it and yes, my sim is now running very well indeed with just the inclusion of the AM setting fixing the problem.