...by Rightmost LP I mean in the binary representation and going by whatever app used to show LPs active the Task Manager displays are LP zero is Top Left..

...that's a little cryptic:

What I'm saying is that you need to look for advantage with certain tasks that when kept to a core without sharing out the sister LP perform best. But that's the trick right?

You mention that cores are maxed or near to max in your tests. That's just not good enough when trying to see what you are looking for - even though you will max it all for the flight, that's 'using', not 'testing for data'. Near max will be flatlining something within the sim fogging your view.

So you would have to reduce settings (or increase) until you can see how the occupied LPs settle out. Some don't do much at all except when loading, but some seem to continue with some processing at all times for example the main task on the first LP.

It's up to you to locate those with choosing the AM and assign your addons away from those for best results.

Remember the only reason to gang up pairs is because they ONLY load the scene and do not require uninterrupted throughput. I hope that helps.

I love threads like this, gets my head up to speed in the morning without having to use coffee 😉

Thanks Mike and Steve!

Great thread, thanks for taking the time to do this. Looks like I’ll be spending bank holiday Monday playing with SMT on/off and affinity mask’s! 🤗

Guys, remember though it's not a great deal maybe 6%, but each few percent gives the edge to the feeling we can get from the sim.

Gerard, that's right about heat because HT will intensify work on the paired cores. So always be careful with the overclock, make sure you are down the heat curve before making changes to the system. P3D can really warm things up when loading the sim.

Many cores systems might not have to enable many or any dual LPs for the background tasks simply because there's plenty of cores. However excess tasks don't add performance they thin out the shared cache and reduce performance into more heat.

Edited August 27, 20187 yr by SteveW

Ive got  my 8700K running using  00,11,11,11,01,01 @5Ghz max temps on any core at any time 75C, Ive just reinstalled P3D clean and at stock with default a/c it can run 50fps locked quite happily.  If I can maintain 30fps locked with complex a/c now I'm a happy camper :)

I don't think I'll be able to resist playing a little more with slightly different AM core pairings though 😛

Edited August 27, 20187 yr by kevinfirth

18 hours ago, kevinfirth said:

Interesting stats Mike, but surely what this shows is only which AM results in the lowest scenery loading time.  The more important data would be performance in sim for each of these AMs re frame-times?  Im not sure that AMs good for scenery loading are necessarily just as good for flying..?

Hi Kevin,

You could be right. I believe it’s simply a case of my needing to start somewhere without over complicating this process. Fast loading of scenery is important and will likely have an impact as regards frame times, at least that’s how I see it. Testing thus far seems to indicate that this might be true. However, like many others around here, I’m still trying to get my head around this somewhat tricky issue.

Steve’s expert input is a huge help. I think we can all agree that his understanding of the nuts and bolts is exemplary, but oftentimes the language employed to elucidate arcane concepts to dunderheads like myself is not always easy to follow and, consequently, implement correctly. Under such challenging circumstances his patience and steadfast willingness to engage with us is truly extraordinary and I, for one, will always welcome his valuable contributions.

Regards,

Mike

Don't be too hard on yourself if you struggle with this tricky subject. I see Kev's set-up looks very expertly done.

@Kev, don't be disappointed if you can't drag any more from it as the returns rapidly diminish if you are preserving the right tasks already.

Remember that it's where our addon exe apps fit in, be sure to give those at least two LPs minimum even putting on the same core if necessary (something you can't do with HT off).

An aggressive app on the wrong LP will cause stutter - even if that LP is only doing 30% if it does not want to be interrupted. This is one of the reasons for such varied results and perplexity when attempting to resolve issues by adjusting the AM of the sim - that's not going to work.

We set up the proper AM and then we leave it alone and get on with problem solving the rest.

 

 

1 hour ago, kevinfirth said:

Ive got  my 8700K running using  00,11,11,11,01,01 @5Ghz max temps on any core at any time 75C, Ive just reinstalled P3D clean and at stock with default a/c it can run 50fps locked quite happily.  If I can maintain 30fps locked with complex a/c now I'm a happy camper 🙂

I don't think I'll be able to resist playing a little more with slightly different AM core pairings though 😛

Hi Kevin,

Regarding temperature my experience has been somewhat similar. Using my ‘new’ Affinity Mask I flew around for an hour or so after taking off from ORBX’s Darrington (1S2) while monitoring core temps throughout the flight. The max temp observed on any one core was 71C. Most of the time temps were much lower.

Regards,

Mike

18 hours ago, newtie said:

Agreed. I'm not sure how this relates to anything except loading times?

Basically it should be obvious to you - the faster the sim can load, the faster your scenery loads and the better inherent performance the simulation has overall.

THIS is the core reason for performance. fps is simply down to your hardware specification and sim settings.

Talking about "fps gains" is - as I keep on saying  < A GIGANTIC WASTE OF TIME> lol

If you want fps get a faster card or turn down settings screen resolution etc.

37 minutes ago, kevinfirth said:

Ive got  my 8700K running using  00,11,11,11,01,01 @5Ghz max temps on any core at any time 75C, Ive just reinstalled P3D clean and at stock with default a/c it can run 50fps locked quite happily.  If I can maintain 30fps locked with complex a/c now I'm a happy camper 🙂

I don't think I'll be able to resist playing a little more with slightly different AM core pairings though 😛

A note to others when attempting to understand the performance characteristics of the sim in any way it is preferred to be done:

Like Kev here; ensure you know what the performance is like without complex scenery a/c and dlls - bring those in and check carefully as they can put a dent in the flow of the sim that just won't shrink like the way improving underlying performance of the loading and data assimilation aspects does it.

Let me try and expand on that addon/dll situation:

 

Take a dll or special gauge .gau for example. Often they invoke SimConnect clients, but these processes and the SimConnect client are run by the sim within the sim AM that is if they are dlls, the exe apps run on their own but also create SimConnect clients that expand the use of the sim within those cores of the sim those allocated by the AM. Remember that even "no-AM" is a fully fledged (highly dodgy - lol) AM, so for example no-AM on a four core is AM=15 tut-tut.

So to add in a dll or complex plane, can and will add an overhead that blurs what we are looking for, namely an underlying setup that supports the best data bandwidth.

THEN we can add in that stuff with confidence.

 

Expansion within the cores initially allocated to the sim can be alleviated in the way FSX boxed works - in that after the layout of the main tasks the entire CPU is allocated for expanding tasks, systems resources and spontaneous threads (expanded affinity). P3D doesn't do that FSX thing because going forward (now) we have 'many cores' and that would not be appropriate, instead we can with appropriate software allow extra cores after the layout but in general this is too advanced a technique at this stage.

 

 

Guys, thx for all this expert input, highly appreciated. But don't forget, the majority of user don't have 8 cores or more up the sleeves, i guess most of us still struggle with 4 cores or even less 😉

1 hour ago, Cruachan said:

You could be right. I believe it’s simply a case of my needing to start somewhere without over complicating this process. Fast loading of scenery is important and will likely have an impact as regards frame times, at least that’s how I see it. Testing thus far seems to indicate that this might be true. However, like many others around here, I’m still trying to get my head around this somewhat tricky issue.

Scenery loading times are important to some people I agree.  I'd be interested to see in sim metrics using those same AMs to see whether AMs good for scenery loading times are also the best for actual flying scenarios as well.  If so, then your methodology would probably provide a relatively easy test procedure for people to set their sims up almost optimally without needing a flying scenario...  (sorry Steve, I know that might leave your tool less necessary 😛 )

1 hour ago, Cruachan said:

Steve’s expert input is a huge help. I think we can all agree that his understanding of the nuts and bolts is exemplary, but oftentimes the language employed to elucidate arcane concepts to dunderheads like myself is not always easy to follow and, consequently, implement correctly. Under such challenging circumstances his patience and steadfast willingness to engage with us is truly extraordinary and I, for one, will always welcome his valuable contributions.

+1,000,000  (Fellow dunderhead here!)

54 minutes ago, SteveW said:

I see Kev's set-up looks very expertly done.

Only because you helped me with it 😛

54 minutes ago, SteveW said:

@Kev, don't be disappointed if you can't drag any more from it as the returns rapidly diminish if you are preserving the right tasks already.

From what I've learned from you the first two tasks separated out onto 01,01 are the most critical...

55 minutes ago, SteveW said:

An aggressive app on the wrong LP will cause stutter - even if that LP is only doing 30% if it does not want to be interrupted. This is one of the reasons for such varied results and perplexity when attempting to resolve issues by adjusting the AM of the sim - that's not going to work.

I haven't put in any addons yet, they will all be reserved for core 6, the 00, we'll see what happens then!  Or do you think I'd gain anything by making the AM 01,01,11,11,01,01 and splitting the addon exe's over two LP's on separate cores?

 

19 minutes ago, kevinfirth said:

Scenery loading times are important to some people I agree.  I'd be interested to see in sim metrics using those same AMs to see whether AMs good for scenery loading times are also the best for actual flying scenarios as well.  If so, then your methodology would probably provide a relatively easy test procedure for people to set their sims up almost optimally without needing a flying scenario...  (sorry Steve, I know that might leave your tool less necessary 😛 )

+1,000,000  (Fellow dunderhead here!)

Only because you helped me with it 😛

From what I've learned from you the first two tasks separated out onto 01,01 are the most critical...

I haven't put in any addons yet, they will all be reserved for core 6, the 00, we'll see what happens then!  Or do you think I'd gain anything by making the AM 01,01,11,11,01,01 and splitting the addon exe's over two LP's on separate cores?

You can put addon exes onto cores doing background tasks of the sim as well, but this slows down that data bandwidth - an aggressive app can bring down a task occasionally. These are tasks that take seconds to complete so are 'relatively' unaffected. If you need to do that you already have confidence in that you started with a good bandwidth setup.

Don't be so modest Kev, you knew full well the techniques we discussed, maybe you just needed a leg up to get on track quickly, no harm in that.