Here's a thing to try with P3D. With way too high settings to crush the performance, try flying a jet at max after burner to the next town. Fast enough to get there so that the scenery is still loading up where we started. Pause and see if any of the fill-in is still appearing. This can become blocked and the fill-in is stalled or runs very slow. While that is going, check or uncheck a box in set affinity and if the fill-in bursts into life there's an issue. What FSX does is Expand Affinity after the layout which is like checking a box in TM details set affinity.

Say with a six core CPU as a purely example situation: We give FSX an AM 011110. When we check in TM we see that all are checked 111111.

In P3D with the AM set instead to use 001111 rather than 011110 so that core zero is used, the threads remain active filling in the scene. That's something that can be checked on a per system basis.

We might not use every available core or LP for P3D because when the loading starts in earnest this can suck resources from the main task. If we leave a core free that will attract those resource requirements away from the sim cores.

A balance is required, especially on many cored CPUs.

An important aspect of micro stutter in P3D is that when VSync is enabled, early frames are ignored. Consequently there is a larger period than usual to get the next showable frame up. So I see some limiting at or below the target rate of the monitor. That's the best thing to do since limiting at or above the target frequency loses frames. Frames arriving too fast are ignored.

One scenario is that the settings are high enough to bring the fps down below the target frequency all the time. In that case we can use VSync=Off.

Another scenario is to use the fps lock slider. Now try VSync=Off compared to VSync=On.

The VSync=Off variation continues to display the next frame without timing to VSync. Parts of the scene may appear shifted as we look around, but we maintain a smoothness. Setting VSync=On can result in tiny notches as it syncs each frame.

These kind of things can be exacerbated when fps is increased slightly and the sim can looks worse. So it is important to know the details of stutter before deciding that one way in particular may be the highway.

So although VSync=On is always recommended, there are times that might not be what we want and it can cause micro stutter.

 

 

This again.

It's a placebo. Placebo can sometimes work too, so... 🤔

18 hours ago, Bert Pieke said:

My monitor only runs at 60 Hz, so yes, for me, that is the reason.

If you can run your monitor at 30 Hz, that is likely better..

My monitor was locked to 60Hz as well 

@4096X2160   - 60 hz

@3840x2160    -30 hz

 

So changing my resolution ever so slowly allowed me to  also change from 60hz to 30hz

I don't use AM or outside tweaks and P3DV4.5 runs on all cores.

https://www.flickr.com/photos/154944674@N05/49166722076/in/dateposted-public/

HT off 4K.

Edited February 18, 20206 yr by rjfry

I'm on 11,11,11,11,11,11,01,01,01,01 =1048405 18 core HT enabled

I’m on 11,11,11,11,01,01 = 4085  8 core HT enabled. Last 2 cores (4LPs) reserved for various Addons. Have determined P3D works optimally on my i7-5960X using 6 cores.

Mike

Yay Mike!

Thanks for your input seems to fit the theory that those HT Pairs, '11's, can pull in the scenery a little more quickly, fps more or less the same.

 

Increasing core counts becomes more interesting.

Having said that, the four core is an interesting example.

HT disabled we have 1111 so we can't do much if anything with that. Coral addon exes to 1110 so as to avoid core zero.

HT On equivalent is 01,01,01,01 this should perform the same as HT Off. Corral addon exes to 10,10,10,00 and this holds our main task on song.

 

Underlying multithreaded resources like file and network are accelerated through optimised cores since they use 11,11,11,11 this should register a tiny bit.

 

So in theory we can combine LPs on those three cores and not affect the main task 11,11,11,01 and in this way load up the scene faster. With the same core speed and settings the fps is the same.

The problem is that one LP can attract more work than what we desire 00,00,00,10 the sister LP of core zero that shares with the main task LP.

We can try 01,11,11,01 which attracts more work to 10,00,00,00

 

We are not talking about much differences because there's not much to arrange. We may find that's not so good as HT Off overall. And with HT enabled our other games may suffer. So it's an expert choice.

 

Edited February 18, 20206 yr by SteveW

25 minutes ago, SteveW said:

And with HT enabled our other games may suffer.

Not been my experience, Steve. I have quite a number installed on this, my ‘gaming’ rig, and all seem to be running very well without the need for any AM interventions. Monitoring core temps confirms they rarely rise above 76C during periods of heavy load.

Regards,

Mike

Edited February 18, 20206 yr by Cruachan

That's good to hear Mike. In theory they should. Those other games generally don't run with any kind off specific tasks per core, that's up to the jobscheduler. Also they are often designed to fit on the console and run within a specification. P3D and FSX being very different in that there's always an AM that lays out one task per LP.

15 hours ago, SteveW said:

No answer!

Tough crowd.

 

My brief trial instances where I checked the above suggestion were: AM removed from cfg, 85 (my usual AM setting that I found to be relatively “smoothest” and use in my day-to-day simming) 84, 170, 255, and 254.

 

Vsync on and off in-sim (I default to on), triple buffering off.  144 Hz monitor running at 120 Hz  with 40 FPS locked in sim.

9 hours ago, CaptKornDog said:

Tough crowd.

 

My brief trial instances where I checked the above suggestion were: AM removed from cfg, 85 (my usual AM setting that I found to be relatively “smoothest” and use in my day-to-day simming) 84, 170, 255, and 254.

 

Vsync on and off in-sim (I default to on), triple buffering off.  144 Hz monitor running at 120 Hz  with 40 FPS locked in sim.

85 (or 170) being four LPs one per core, so that should work the same as HT Off with the benefit of a tiny lift from system resources having 8LPs 2 per core in HT mode. stick to 85 with LP zero used.

255 is no AM is AM=0 in HT mode 11,11,11,11

254 is '10' on the right.

Instead try 253 11,11,11,01.

Referring to my comment above 'increasing core counts', 29 would be worth a try 01,11,11,01.

So I would try 253 and 29.

If I have no AM and switch in HT from no HT then I have a lumpy sim, 18 core, instead I use ten cores with 6 paralleled

11,11,11,11,11,11,01,01,01,01 = 1048405

My point about 'no answer' is that results of any tests of this nature depend entirely on the way the system is set beforehand.

The usual comments related to HT enabling but with no AM means a fail right from the start

No AM in HT puts the simulator on the first two cores.

The minimum requirement of HT is an AM 85 01,01,01,01 as a starting point that continues the test with the sim at one LP per core as it was in HT off,

On 2/17/2020 at 12:47 AM, Bert Pieke said:

And what I would add is that with Nvidia driver 442.19, you can limit your fps in the Nvidia Control Panel (I have mine set to 29..) and this will cut back on your Core 0 usage..

So, now I have fps set to unlimited in P3D, vsynch off, fps capped at 29 in NCP, and I get smooth flight, and average Core 0 load of 60-70%..

This with a 1080p/60Hz monitor.

 

Bert, I don't get this. Ok, you can limit your fps in NCP to, say, 29 or 30 frames, vsync off, but your monitor is still at 60Hz. I understand therefore they are not synchronized yet them your game and your monitor. See if adding Vertical sync Adaptive (half refresh rate) in NCP draws an even better result.

Edited February 19, 20206 yr by Dirk98

Is Low Latency Mode in NCP of any good for Prepar3d?

Thanks.

Edited February 19, 20206 yr by Dirk98

For me, P3D v4.5HF2 could be the best sim ever created if the dev team would address and fix the occasional pauses that occur when approaching airports with AI & the loading AG in the distance. Aside from those two annoying things, P3D v4.5 is as good as it gets.

I'm only saying this because I have a loaded sim with beautiful photo sceneries, beautiful night lighting, lots of airports, AI, ASP4, SkyForce, the right airplanes etc. Without add-ons, there probably would be any issues, but P3D v4 is only as good as the add-ons that are installed.