NOTAM...any Affinity Mask setting in your .cfg could be actually counter-productive!

[Sesquashtoo 2,846]

I wish there was a way to edit thread titles when you typed a typo!  There should be....jeez...lol.

------------------------------------------------------------------------------

 

I have been running with an AM mask of either 233 or as SteveW suggested, a better setting for my H.T. mode of 116.

 

Well this morning I decided to remove the AM .cfg entry in its entirety.  I then set my self up at LatinVFR's  Key West v2.

 

What I saw in the FPS 'test' department, and operation smoothness, was nothing less than amazing.

 

Well...here are screen captures of my taxi, rotation, and fly around...with the displayed FPS readings, that can go no higher than 33 FPS locked within the sim.

 

I was getting higher FPS readings than with an AM entry in place.  

 

I then went to a very congested LatinVFR Miami v3...and the same deal...actually better FPS performance.  I only now talk FPS for the typing-point of being able to say, something is 'greater' than it was with this or that placed in the .cfg file.

 

Bottom line....try removing (if you have an AM entry in any of your MS. based flight sims...) the Affinity Mask entry, and then go fly at a familiar haunt.  See if you FPS and smoothness performance is equal, or in fact actually better with no AM entry at all.....

 

I was using an AM entry to support my 4 LP 'cores'...but apparently, one is not needed even if you run in Hyperthreading Mode.  Interesting....at least for use within P3D v3.2.3!

 

Ses

-----------------------------------------------------------------------------------------------------

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Notice that reading of outside temp...82F!!!!!   Man, I wish today this was more than 'virtual'....drab around the Great Lakes today...cold and drab....

 

 

Here's two shots within ORBX Southern California, which is deadly to FPS, even if the sim is dumbed down...but it isn't here..and take a look at those FPS readings flying near Cold Water Canyon, on my way to the coast...no Affinity Mask in the .cfg file...and most impressive!

 

 

 

 

Non AM coastal shot near Point Magu:

 

 

 

 

 

 

[TheBoom 31]

It could be that with GA the threads are better optimized with no AM. Try that with a tubeliner I'm pretty certain it will be the opposite. But either way, to each his own.

[Sesquashtoo 2,846]

It could be that with GA the threads are better optimized with no AM. Try that with a tubeliner I'm pretty certain it will be the opposite. But either way, to each his own.

Boom, I don't believe it is dependent on what type of aircraft you have...it is sim-base dependent.

 

P3D v3.2.3 is most likely already optimized for what job, should be on what thread, right out of the box....

Adding a user-AM probably throws a wrench into the gears...lol.

[jrs182p 23]

Interesting, but I don't buy it either. After my extensive testing on my system, using no AM with HT on results in a lower frame rate and greater VAS usage. Each to their own I guess! Your tests really don't seem to be that accurate. For a start your not using the exact same scenario for each trial. The change in weather/clouds alone can change your FPS.

[Pete Dowson 558]

Surely it is dependent upon whether you run any other processes on the same PC. Obviously (?) it is not really a good idea to restrict P3D's use of processors when there's no need.

 

The use of AM settings really only comes into its own when there are other processes running which could conflict. You then need to split them onto different processors.

 

Pete

[Midnight Music 252]

Agreed, Pete! My i7 4770 with HT on and a 980ti runs much better in V3.2 with no AM setting. The only tweak I use us FFTF at 0.10 I use tubeliners all the time and have all the usual addons and I even use NVI with 4Xssga for a great look. I keep my sliders in the half to 3/4 range. All good.

[Sesquashtoo 2,846]

Interesting, but I don't buy it either. After my extensive testing on my system, using no AM with HT on results in a lower frame rate and greater VAS usage. Each to their own I guess! Your tests really don't seem to be that accurate. For a start your not using the exact same scenario for each trial. The change in weather/clouds alone can change your FPS.

I did two heavies...and both were consistent high FPS.  

Surely it is dependent upon whether you run any other processes on the same PC. Obviously (?) it is not really a good idea to restrict P3D's use of processors when there's no need.

 

The use of AM settings really only comes into its own when there are other processes running which could conflict. You then need to split them onto different processors.

 

Pete

Well..it sure seems so on my rig.  I had been using AM all along..thinking it better.   Well....it seems to be 'better' with a clean .cfg file...lol.   I'm getting great results with no AM setting in HT mode.

 

Mitch

[glider1 90]

Surely it is dependent upon whether you run any other processes on the same PC. Obviously (?) it is not really a good idea to restrict P3D's use of processors when there's no need.

 

The use of AM settings really only comes into its own when there are other processes running which could conflict. You then need to split them onto different processors.

 

Pete

 

Could not have said it better. AM is not needed without addons. It is needed when addons are using CPU, interfering with the main sim job which causes random stutters. The problem is that at the moment it appears that AM in P3D causes issues with blurries because the rendering threads seem to get stuck sometimes, unable to finish back-filling the textures.

[TheBoom 31]

Could not have said it better. AM is not needed without addons. It is needed when addons are using CPU, interfering with the main sim job which causes random stutters. The problem is that at the moment it appears that AM in P3D causes issues with blurries because the rendering threads seem to get stuck sometimes, unable to finish back-filling the textures.

So have you tested it wIth no AM? I thought that issue occurs regardless of whether you set an AM or not.

 

Edit: To answer my own question. Just tested it, no AM HT OFF, blurries still present.

[SteveW 1,792]

Remember that if we run Prepar3D (P3D) on a Hyper Threading (HT) Central Processing Unit (CPU) with the default Affinity Mask (AM) of zero, and be that CPU a dual core, or a many core, or even a very many core, the first four P3D threads will occupy the first four Logical Processors (LP) and thus the entire simulation rendering system will be running on a dual core CPU. When the sim is loading scenery at the loading progress window, the third and fourth threads will be fully utilised. The other LPs of how ever many remaining cores are available on the CPU are utilised to improve that scenery loading time.

 

 

The good news is that P3D runs very well on a dual core CPU, the bad news is that four cores, HT or not, doesn't provide scope for many configurations. With four core CPU's HT On and no addons AM=85 or AM=170 provides the best performance. Going down to three cores, to give one up for addons, AM=116 works best because the second thread is idle while the third and fourth threads collect scenery. With six cores, the scenery loads faster, but the stability of the sim suffers as the main job and thread pool have more work to do with the extra threads.

 

 

But jolly good show! From reading the posts above it seems that what's finally sinking in is that Hyperthreading provides better performance unless we're running a very well prepped overclock.

 

 

Regarding changing priority; remember that if we give P3D a higher priority, the sub-processes it relies on have to wait. So elevated sim priority will be blocking the very stuff it's waiting for. Also not good, lower priority events get queued up and fire off all at once later on. There's no threading bug in P3D.

[Egbert Drenth 250]

I'm still dazzled by the (correct) AM setting.
Have experimented briefly and some settings work OK, others do not (like 85)

Currently I'm running 238 on a 6700K with HT on. Why 238? I don't have a clue  :smile:

 

My confusion starts with the use of add-ons...
What defines an add-on? Is it an external application or also the ones that run internally in P3D, like FS2Crew, FSUIPC ?

I use 2 networked systems:

• Main system running P3D (+ vPilot proxy and EFB dataprovider)
• My 2nd system runs add-ons like WideFS, ASN, vPilot client etc.

 

So what would be the best AM for this specific system?
Are vPilot proxy + EFB DP considered as add-ons or doesn't is really matter?

 

[glider1 90]

What defines an add-on?

 

Hi Egbert

The definition of an addon means any executable file that is capable of running on separate cores (called LP with HT on) to the simulators own internal cores (LP). There are internal addons that cannot be forced off the simulators default cores and nothing can be done about them. These are the addons that launch from the dll.xml file. External addons are launched from the exe.xml file or from the operating system itself or run on the network. External addons can be isolated from the simulators own internal cores.

 

Are you having problems? If not, don't bother with an AM. It is used in attempts to reduce stuttering. Stuttering can be caused by a number of things. If you are sure that there is no other explanation for stuttering other than your external addons causing it, then start thinking about what AM to go with. There are plenty of threads on that.

 

EDIT: I should mention that other programs running in the background can also interfere with the simulator, which is another reason why AM is used.

[hesynergy 67]

Could not have said it better. AM is not needed without addons.

 

And how many of those of us, do you suppose, are flying without add-ons?

 

Cheers to an imaginary world,

 

Chas

[GSalden 4,655]

On my 5820K ( 6 cores ) I tried without AM and with AM.

With AM, with help from Steve, I tried 1360, 340 and 1396,because I am running several cockpit addons.

Sometimes 1396 seemed better in heavy scenery but then after a while the framerate started to drop below 10..

Finale AM 350 icw Proces Lasso for assigning the addons to remaining cores is the best combo for my system

[Egbert Drenth 250]

Are you having problems? If not, don't bother with an AM. It is used in attempts to reduce stuttering.

 

Hi Glider,

 

Thanks for your explanation.

The reason why I started to experiment is that I have (at airports) tiny mini-stutters.

The stutters are not really a nuisance, but if possible I would like to get rid of them.

So far setting an AM didn't really solve the issue compared to no AM.

But what I understand from other replies, it that it is just a matter of trying which AM setting fits the best