Hyperthreading off works great!

[Sartanius 399]

11 minutes ago, kevinfirth said:

Not all cores should necessarily be loaded in a balanced manner.

The primary task is the renderer, should aim for that to be 90-95% loaded, masking off the virtual core if using HT.

The second task does a lot of the managing workload, Steve recommended that be kept also to a physical core by itself.

The rest is scenery loading which will do work as and when necessary.

Anything that impact on the primary or secondary tasks will limit fps and/or induce long frames (aka stutters)

Thanks for responding, Kevin.  Help me understand this.  If all cores are running with balanced loads, and no one core is pegged at 100%, then could one say that there's no bottleneck since each core still has headroom left to do more?  If that is the case, then every thread is getting it's turn and doesn't have to wait.  How can the processor service threads any faster than that, assuming constant clock speeds?

[kevinfirth 1,392]

8 minutes ago, dmiannay said:

Thanks for responding, Kevin.  Help me understand this.  If all cores are running with balanced loads, and no one core is pegged at 100%, then could one say that there's no bottleneck since each core still has headroom left to do more?  If that is the case, then every thread is getting it's turn and doesn't have to wait.  How can the processor service threads any faster than that, assuming constant clock speeds?

Not all tasks are created equal.

Primary task renders the scene - this needs access to a core unfettered by any other process.  Impeding this task will limit your framerate and probably introduce unmanageable stutters.

Secondary task manages a lot of the other stuff.  Its not as critical as the renderer but if you can give it a core by itself that will reduce the opportunity for conflict and performance degradation.

The scenery loading tasks you can let do their thing whatever.

If you don't run HT on then you don't need to worry.

If you do run HT on you need to definitely mask off the logical processor alongside the core running the rendering task, and if possible the logical core running alongside the secondary task as well.

 

Im running an AM of 1013 - 00,11,11,11,01 ,01 = 1013

Edited December 31, 2018 by kevinfirth

[Sartanius 399]

2 minutes ago, kevinfirth said:

If you don't run HT on then you don't need to worry.

If you do run HT on you need to definitely mask off the logical processor alongside the core running the rendering task, and if possible the logical core running alongside the secondary task as well.

Ok, I get it now.  Thanks for your explanation, and Happy New Year!

[MDFlier 2,031]

24 minutes ago, dmiannay said:

Right now, with an 8700K and HT Off, I have balanced loading across all 6 cores throughout a flight. No one core is more heavily loaded than another.   It is a thing of beauty to see. I don’t run any AM, just a stock configuration overclocked to 5.0GHz.

P3Dv4.4 working this way out-of-the box is what we’d all been hoping for, right? What am I missing here?

Exactly the same configuration I'm using with my 8600K (which I believe is an 8700K w/o HT, if I'm not mistaken).

I was going to put up a post today about my overall P3D experiences 2 full months after making the switch from FSX. Somehow, I've managed to find a combination of settings that yield a fairly consistent 27 FPS with 4K textures, dynamic lighting, HDR, and 4xSSAA on my i5-8600K, GTX 1060 6GB system. I have only used 2 "tweaks", and I'm not sure if they even count. I added the clipmode=minimum and the adjustment to make opaque shadow textures 4K. For the couple of weeks that I was on 4.3, FPS were in the high teens with the same settings. I really haven't had enough time playing around with things to add anything else useful to this conversation, but I will say that 4.4 runs very well on the 8600K. 

[SunDevil56 183]

15 minutes ago, kevinfirth said:

 

If you do run HT on you need to definitely mask off the logical processor alongside the core running the rendering task, and if possible the logical core running alongside the secondary task as well.

 

Im running an AM of 1013 - 00,11,11,11,01 ,01 = 1013

I'll give that AM a try and see how it compares to my current one.

[Luke 782]

3 hours ago, kevinfirth said:

If you do run HT on you need to definitely mask off the logical processor alongside the core running the rendering task, and if possible the logical core running alongside the secondary task as well.

The part I don't understand is that the Windows kernel is quite aware of which cores are logical cores (since XP) and more importantly has a better understanding of all the threads in the machine and when they need to execute.

Cheers!

[kevinfirth 1,392]

11 minutes ago, Luke said:

The part I don't understand is that the Windows kernel is quite aware of which cores are logical cores (since XP) and more importantly has a better understanding of all the threads in the machine and when they need to execute.

Cheers!

Does it really even fully with P3D? My 'knowledge' stems from @SteveW, and I trust his professional opinion and the integrity of his test data.

Edited December 31, 2018 by kevinfirth

[SunDevil56 183]

Kevin, I run HT "on" because that's what has given me the best results, and I tried your AM of 1013.

While I can't say it's giant improvement on my 8700K/1080 Ti 11gb system, I do think it made things ever so slightly smoother, and reduced micro stutters.

Not that I was having trouble with smoothness or stutters with my old AM, but every little bit helps, so thanks for the tip... 😎

[pegruder 65]

I haven't really been having a micro stutter issue but I may try disabling HT on my i9 9900k and see if there's any improvement in performance.  My frame rates certainly bounce all over the place for sure.  Some cities (NYC - always a problem area) with the FSL bus I see a range of fps from 12-40 as I'm flying around.  I be clear - I dont mean if I look one way its 12, the other 40, i mean the fps counter is going nuts and fluctuation that much even If Im just sitting still.  Ill have to give the HT off a go and see what happens.

[Luke 782]

4 hours ago, kevinfirth said:

Does it really even fully with P3D? My 'knowledge' stems from @SteveW, and I trust his professional opinion and the integrity of his test data.

Prepar3D is just another user-mode Windows process. Why wouldn't it?

Cheers!

[kevinfirth 1,392]

1 minute ago, Luke said:

Prepar3D is just another user-mode Windows process. Why wouldn't it?

Cheers!

A professional dev very experienced in the intricacies of flight sim software architecture, with proven test data backing up his advice...v why wouldnt it? 

You'll need to do better than that Luke sorry, I have never done blind faith, especially in MS!  I'll go with the expert and test data until there's something that convinces me otherwise 😛

[Bert Pieke 7,584]

37 minutes ago, Luke said:

Prepar3D is just another user-mode Windows process. Why wouldn't it?

Cheers!

LM has clearly taken steps to distribute the P3D workload across multiple cores in a very deliberative way.

The OS has no clue what is running and is not going to "improve" anything..

Each user can try to adapt the P3D workload based on their specific CPU configuration (number of cores, OC, HT on or off etc) but that may just possibly make things worse, not better 

Some users go into Task Manager and re-enable the Affinity settings after P3D is loaded.. that changes P3D behavior.. probably by undoing something LM intended.. is that better?

It may look like the workload is more evenly shared across the cores.. but is that better?

You can never stop users from fiddling with their settings, and bragging about their findings on the forums, but that does not mean it applies to the population in general.. I kind of think that LM used some smarts when they designed V4.4. the way they did..

Edited December 31, 2018 by Bert Pieke

[Mace 1,902]

34 minutes ago, kevinfirth said:

A professional dev very experienced in the intricacies of flight sim software architecture, with proven test data backing up his advice...v why wouldnt it? 😛

His testing and comments were with 4.3.  

I have yet to see him say anything as to 4.4.  What is assured, is that 4.4 behaves differently as to texture loading/ cpu usage. 

[kevinfirth 1,392]

3 minutes ago, Mace said:

His testing and comments were with 4.3.  

I have yet to see him say anything as to 4.4.  What is assured, is that 4.4 behaves differently as to texture loading/ cpu usage. 

My recent posts have been based on personal conversations post 4.4 and so are very applicable..  just because it's not been said publicly doesnt mean its not true :)

[Mace 1,902]

1 minute ago, kevinfirth said:

My recent posts have been based on personal conversations post 4.4 and so are very applicable..  just because it's not been said publicly doesnt mean its not true :)

I see no one in this thread denying any truth or fact.

What is also fact is that 4.4 is acting differently than 4.3 as to the matter at hand.