Why is P3D handling VAS so poorly?

[threexgreen 3,139]

On 5/14/2017 at 1:00 PM, tooting said:

That's great.  It works for some cards bit not all.  I can promise you with a 980ti and an Asus board it doesn't.

Like Baron said each system is different.

For example on mine and my mates Asus boards if you remove a USB joystick or pedals by mistake by knocking it while p3d is running it ctds.  Now you could argue until the cows come home thats bad coding, but p3d would blame windows and windows would blame Lm.  So the problem never gets fixed.  Same with these drivers.   No ones going to claim responsibly for it.  It costs time and money 

I have a 980ti and an ASUS board and it works just fine. I can also remove whatever I want and P3D doesn't care what I did.

[Ray Proudfoot 8,952]

On 12/05/2017 at 5:30 PM, SteveW said:

Give addons at least two LPs. Here's an overview of setups for a 4core+HT using three and four cores:

The four core versions:
01,01,01,01 = 85 dec - sim - best rendering
10,10,00,00 = A0 hex - apps (for batch file setting)

 

After a couple of days of testing with various scenery settings I've now arrived at a point where I'm 85% happy with performance versus appearance versus VAS. As suggested by Steve above I'm now using an AM of 85 together with A0 for the 3rd party programs that run alongside P3D. That's easily setup with SimStarterNG, a great piece of software.

My test flight from Oslo to Heathrow resulted in 550Mb of VAS remaining after I parked Concorde at T4. More than happy with that.

I said 85% and not 100% because the LOD Radius needs to extend further as it's possible to see the boundary when you're cruising at FL500 and above. But there's very little difference between Ultra and the top setting and it does eat into VAS. Ultra is the best compromise.

The other thing I need to look at is Cloud Draw distance. At FL500 and above the horizon is a couple of hundred miles away and a distance of 80-90 miles for clouds to appear is much too close. Anyone got any thoughts about the ideal distance without VAS being gobbled up?

Coming into land at EGLL with frame rates around 25-30 is great. Ai is set to 150 but I could do with prioritising the gates as Ai seem to park in all the wrong places. Apart from these fairly minor things I'm really happy with P3D.

[Sumits81 9]

On 5/12/2017 at 4:28 PM, SteveW said:

Hi, variants based on the 116 pattern for three core use is good because it utilises three cores more efficiently than any other setting. 85 or 170 (same pattern) will provide better performance since they use four cores - but not by much. With the three core variants it leaves a core free for addon exe apps. With four cores used we have to be more careful with the way these apps layout across the CPU, keep them on the last two cores whereby they share cores running background tasks only. In Rays examples he appears not to have tried 85 or 170 HT on for some reason but even so would not give any indication of true performance. Those results show a very obvious change in fps across the board which suggests problems associated with other processes are disrupting the flow of the sim and the AMs produce different results around that behaviour, that is all. They don't show throughput capacity since it's not linked directly to fps. 

Hi Steve,

Once again, thanks for all your explanations and suggestions. With AM116 on my i7-6700k, i noticed blurries on ground textures (Orbx Global, EU LC + Vector). After a while of t'shooting I've narrowed this down to the AM and have now changed it to 85 instead. Textures appear to be loading fine and no blurries so far. 

Cheers

[SteveW 1,792]

116 uses four LPs of three cores and part of the background process is shared on a core so the only reason to use 116 is when we need to allocate more bandwidth to addon exe via core zero and it's better than using 84 or 168 or with HT disabled 14 since these use three LPs putting even more stress on the rendering stage. 85 gives all four LPs a core to themselves. Beware that obvious visual problems as you describe are not down to the differences between these two AMs but rather the interaction of the sim with other processes. In a perfect system you will find it very hard to see the difference between 116 and 85 used, so your blurriness problem is most likely another aspect was altered by the change in AM.

[Sumits81 9]

43 minutes ago, SteveW said:

116 uses four LPs of three cores and part of the background process is shared on a core so the only reason to use 116 is when we need to allocate more bandwidth to addon exe via core zero and it's better than using 84 or 168 or with HT disabled 14 since these use three LPs putting even more stress on the rendering stage. 85 gives all four LPs a core to themselves. Beware that obvious visual problems as you describe are not down to the differences between these two AMs but rather the interaction of the sim with other processes. In a perfect system you will find it very hard to see the difference between 116 and 85 used, so your blurriness problem is most likely another aspect was altered by the change in AM.

Thanks for that. It all makes sense. 

I was using AM116 for the longest time without any issues and started experiencing blurries in the last few days. I've purchased Chaseplane, PrecipitFX, A2A C182 and the Majestic Dash8 in the last few days and it may be that the additional load is causing these blurries (Chaseplane perhaps?).

The blurries were pretty bad rendering the ground a mash of green and brown patches. Deleted the prepar3d.cfg file and let it rebuild after which I added the tweaks one by one. That's when I realised AM116 wasn't cutting it and AM85 did.

Any thoughts to why 116 no longer did the trick?

Thanks

[SteveW 1,792]

Put very simply, with 01,11,01,00=116 Let's assume bigger addons and new activity starting on core zero. Since other processes gravitate to the locations of the other zeros means there's a bigger impact from activity moving onto LP7 - only one core with a sim background process, whereas with 85=01,01,01,01 the load is distributed onto LP7 and LP5 - two cores.

[asanal 32]

Steve, Hi

I have i7-5960X CPU@3GHZ, 8 Cores with 16 processors.

What will be the best performance settings  for P3D.

AM setting: 21844

HT On

Thanks

 

[Sumits81 9]

4 hours ago, SteveW said:

Put very simply, with 01,11,01,00=116 Let's assume bigger addons and new activity starting on core zero. Since other processes gravitate to the locations of the other zeros means there's a bigger impact from activity moving onto LP7 - only one core with a sim background process, whereas with 85=01,01,01,01 the load is distributed onto LP7 and LP5 - two cores.

Thanks Steve. Understood. 

Is there a way of finding out what services (exe) may be hammering LP0. Looking at task manager didn't yield a great deal of info. 

Cheers

[Cruachan 705]

5 hours ago, asanal said:

Steve, Hi

I have i7-5960X CPU@3GHZ, 8 Cores with 16 processors.

What will be the best performance settings  for P3D.

AM setting: 21844

HT On

Thanks

 

Hi Ahmet,

I suggest you head on over to the 'Codelegend' (Steve's) website (a veritable treasure trove of information and very useful products) and read this comprehensive article:

http://www.codelegend.com/forums/viewtopic.php?f=2&t=532

Within the article you will find a couple of AM suggestions for your CPU. 

You will note from my sig that I prefer to run with HT=OFF so my preferred choice of AM won't be of much help to you, I'm afraid. However, while I was knee deep in testing last year I found the value 21760 to be quite effective. It does no harm to experiment, just keep Steve's basic principles in mind as you go.

Regards,

Mike

 

[SteveW 1,792]

5 hours ago, asanal said:

Steve, Hi

I have i7-5960X CPU@3GHZ, 8 Cores with 16 processors.

What will be the best performance settings  for P3D.

AM setting: 21844

HT On

Thanks

Hi, with HT on that should be fairly good anyway, but instead try 5440 and 5460

 

1 hour ago, Sumits81 said:

Thanks Steve. Understood. 

Is there a way of finding out what services (exe) may be hammering LP0. Looking at task manager didn't yield a great deal of info. 

Cheers

With AM=85=01,01,01,01 the ones represent On LPs core zero-LP0 on the right. In task manager it's the first in the graphs. The way the sims works (in an oversimplified way) is to split into four main processes so one per core would be optimum. We have one major process rendering, supported by an arbitration process handling a pair of data gathering processes. So with your 85 the first LP would always show maximum performance when it can during steady use of the sim, the others show maximum performance when loading a scenario.

For looking at processes try Sysinternals Process Utilities, TechNet, Microsoft - have a look there for process explorer. However, I generally use my own custom built tools to ascertain such information.

[Ray Proudfoot 8,952]

Just completed EGCC-KMCO in Concorde in a little under 4 hours. Weep all you Boeing and Airbus pilots

Anyway, that wasn't the reason I posted. This is...

I was getting concerned about remaining VAS as I flew abeam Orlando passing 10,000ft. I had less than 1000Mb remaining and T2G Orlando has always been one of my troublesome airports in FSX. Anyway, it,or rather P3D passed with flying colours. 800Mb remaining as I taxied off 35R to the terminal. Wonderful!

That spike you see right at the end was when I moved the aircraft from a medium to heavy gate. The VAS increased quite a bit but was then recovered. Good job LM.

[GSalden 4,656]

Glad to see that you have tuned your sim till almost perfection ...

[asanal 32]

Steve and Mike,

Thanks

[Ray Proudfoot 8,952]

1 hour ago, GSalden said:

Glad to see that you have tuned your sim till almost perfection ...

I'm really pleased with the performance of P3D on my 4 year-old PC. I think the Affinity Mask setting has made a big difference. Kudos to Steve for helping there.  The other thing is to keep a sense of perspective when setting the sliders. I obviously had them set far too high originally- hence the title of this thread.

I'd rather have 5-10 fps extra on finals than another 50 buildings showing. 

[Sumits81 9]

Well done, Ray. Great to see a fellow P3D'er doing well in-sim :)

@SteveW I was reading somewhere there may be a better way of setting/assigning cores to P3D via an external software (Project Lasso might have been mentioned). I'm no expert on the subject but the author of this article suggested texture loading was better. Is this a placebo effect or is there any tech-truth to it?

Cheers