Good Framerates - major stutters / pauses

[Christopher Low 5,325]

2 hours ago, ttbq1 said:

Are you sure you dont have NCP? right-click at desktop doesn´t show the option?

Maybe an old Nvidia driver was installed by windows itself?

That's an interesting one, because I also did not have NCP installed after setting up my new Windows 10 PC at the end of January 2020 (with the latest Nvidia graphics drivers available at that time; I do not remember the exact version). However, I recently installed the 446.14 drivers (because I was advised to after the "no NCP" revelation), and this time NCP has been installed. Quite why it was not installed previously will probably remain one of our planet's greatest mysteries.

[SteveW 1,792]

1 hour ago, ShezA said:

I will check it out although I suspect if I use the non-physical cores (1,3) then temps might go up. Thanks.

The way to think about cores is that with HT disabled you have one logical processor per core, that's one virtual core per core. With HT enabled that is two LPs per core, two virtual cores. The HT core is hard to understand. Let's look at one core running two identical tasks. These tasks take a certain time to complete. These tasks are time sliced at a certain frequency so they appear to run at the same time. In order to affect a good result with  HT enabled, those two tasks must be directed one each per LP. When that happens the two tasks finish sooner than they do with HT disabled. Those two tasks are helped to finish sooner because the swapping of the tasks to appear simultaneous is improved. Two tasks can be stored without swapping over their setup and registers (context) each time slice. With the way the simulator works, we can't improve the performance of the main task as it is monolithic. We can improve the performance of the background tasks pulling in the data and constructing the scene by using some of them in pairs on the later cores. If we use those spare LPs for other tasks that too is an improvement overall to the system throughput. We need not increase heat with HT if the setup is organised properly.

[SteveW 1,792]

12 minutes ago, Christopher Low said:

...will probably remain one of our planet's greatest mysteries.

Along with how they manage to stick the non-stick surface to a frying pan.

[ShezA 176]

2 minutes ago, SteveW said:

The way to think about cores is that with HT disabled you have one logical processor per core, that's one virtual core per core. With HT enabled that is two LPs per core, two virtual cores. The HT core is hard to understand. Let's look at one core running two identical tasks. These tasks take a certain time to complete. These tasks are time sliced at a certain frequency so they appear to run at the same time. In order to affect a good result with  HT enabled, those two tasks must be directed one each per LP. When that happens the two tasks finish sooner than they do with HT disabled. Those two tasks are helped to finish sooner because the swapping of the tasks to appear simultaneous is improved. Two tasks can be stored without swapping over their setup and registers (context) each time slice. With the way the simulator works, we can't improve the performance of the main task as it is monolithic. We can improve the performance of the background tasks pulling in the data and constructing the scene by using some of them in pairs on the later cores. If we use those spare LPs for other tasks that too is an improvement overall to the system throughput. We need not increase heat with HT if the setup is organised properly.

Got it. So for me with temp issues the setup settles on

1. P3D on core 0-2,4,6,8,10
2. Other tasks (ASP3D, etal) on 3,5,7,9,11.

Best balance of temp and fluidity. Thanks for the help in further refining my setup. Cheers.

[SteveW 1,792]

11 minutes ago, ShezA said:

Best balance of temp and fluidity. Thanks for the help in further refining my setup. Cheers.

Good post.

[cyyzrwy24 118]

5 hours ago, ttbq1 said:

Are you sure you dont have NCP? right-click at desktop doesn´t show the option?

Maybe an old Nvidia driver was installed by windows itself?

It wasn't there...had to restart....

Yesterday I had way lesser "pauses" and none while in the air.....after 7 Hrs in flight and while landing and at the airport taxiing got it again...not as bad but still noticeable....

[duckbilled 310]

3 hours ago, ShezA said:

Got it. So for me with temp issues the setup settles on

1. P3D on core 0-2,4,6,8,10
2. Other tasks (ASP3D, etal) on 3,5,7,9,11.

Best balance of temp and fluidity. Thanks for the help in further refining my setup. Cheers.

After reading this and another thread, I turned off HT. I had a very smooth flight yesterday, so turning it off seemed to improve things. Before, I was running HT without a AM. Wiled I like the HT off performance, it may be the case that I need a good AM to take advantage of HT. I have to admit, however, I am a little lost after 8 pages. I get the concept, but operationalizing things a little different. Based on your settings Shez, I'm thinking that this may be a good setting to test for my 9900k:

[JOBSCHEDULER]
AffinityMask = 21855

P3D on 0-2, 4, 6, 8, 10, 12, 14

Now, if I do that (or an AM for that matter), how do I assign the other processes to other cores? Is there something else I need to do besides the AM entry? I know this make be obvious to many, but it isn't to me.

Thanks for any guidance on this. 

[SteveW 1,792]

9 minutes ago, duckbilled said:

 

[JOBSCHEDULER]
AffinityMask = 21855

P3D on 0-2, 4, 6, 8, 10, 12, 14

 

That's exactly the kind of setup that will not perform well

101010101011111 = 21855

Instead you want 01 on the right - you  have 11 sharing one core to two major tasks. These are the things that give HT and AMs a bad name. Right across AVSIM.

[ShezA 176]

7 minutes ago, duckbilled said:

After reading this and another thread, I turned off HT. I had a very smooth flight yesterday, so turning it off seemed to improve things. Before, I was running HT without a AM. Wiled I like the HT off performance, it may be the case that I need a good AM to take advantage of HT. I have to admit, however, I am a little lost after 8 pages. I get the concept, but operationalizing things a little different. Based on your settings Shez, I'm thinking that this may be a good setting to test for my 9900k:

[JOBSCHEDULER]
AffinityMask = 21855

P3D on 0-2, 4, 6, 8, 10, 12, 14

Now, if I do that (or an AM for that matter), how do I assign the other processes to other cores? Is there something else I need to do besides the AM entry? I know this make be obvious to many, but it isn't to me.

Thanks for any guidance on this. 

So I use a program called Process Lasso to assign the CPU affinity to the other programs. Its a nice little program which gives a lot of information about the computer and the programs/services currently running. Maybe someone else can advise if its possible to do this without PL.My AM setting actually matches my PL core assignments (I do this in SimStarter P3D). Apologies for introducing these other programs....

[SteveW 1,792]

So long as we use the jobscheduler section of the Prepar3D.cfg for setting the P3D AM we are fine, using managers for programs than manage their own affinity like P3D are better left to do their thing. However programs that can start P3D correctly such as IFPro can also manage the affinity correctly.

Edited June 1, 2020 by SteveW

[Mace 1,896]

1 hour ago, duckbilled said:

Now, if I do that (or an AM for that matter), how do I assign the other processes to other cores? Is there something else I need to do besides the AM entry? I know this make be obvious to many, but it isn't to me.

Thanks for any guidance on this. 

I use a batch file to set affinity for other apps.  I can post it if you want, although mine is for a 6-core HT on setup, and for your 8-core you would have a different affinity.

I played around with Process Lasso with P3Dv4.1 or 4.2 time period, but had poor results.  I prefer using a batch file -- simple -- fast -- it loads the apps and then it closes and it is DONE.  Done as in not in memory any more.

[Ianrivaldosmith 5,317]

1 hour ago, duckbilled said:

After reading this and another thread, I turned off HT. I had a very smooth flight yesterday, so turning it off seemed to improve things. Before, I was running HT without a AM. Wiled I like the HT off performance, it may be the case that I need a good AM to take advantage of HT. I have to admit, however, I am a little lost after 8 pages. I get the concept, but operationalizing things a little different. Based on your settings Shez, I'm thinking that this may be a good setting to test for my 9900k:

[JOBSCHEDULER]
AffinityMask = 21855

P3D on 0-2, 4, 6, 8, 10, 12, 14

Now, if I do that (or an AM for that matter), how do I assign the other processes to other cores? Is there something else I need to do besides the AM entry? I know this make be obvious to many, but it isn't to me.

Thanks for any guidance on this. 

I believe the setting you need for that setup is [JOBSCHEDULER]
AffinityMask = 21845

[duckbilled 310]

3 minutes ago, Ianrivaldosmith said:

I believe the setting you need for that setup is [JOBSCHEDULER]
AffinityMask = 21845

I can try that. That has P3D on 0, 2, 4, 6, 8, 10, 12, and 14. right?

 

[Ianrivaldosmith 5,317]

1 hour ago, duckbilled said:

I can try that. That has P3D on 0, 2, 4, 6, 8, 10, 12, and 14. right?

 

Yeah for an 8 core with HT enabled 

[SteveW 1,792]

That's right! Main thing is to avoid sharing the first core to two P3D tasks which the binary '11' on the right will do.