Prepar3D v5.3 released

[SN737 140]

15 minutes ago, shivers9 said:

It would be helpful if you listed your computer set up.

I did in my post, but here it is again.

9900K @ 5.2 GHz, 32GB ram 3200 MHz and a GTX 1080 Ti

[willy647 193]

On 12/2/2021 at 12:27 PM, Wolkenschreck said:

Interesting.

I noticed that too and first suspected an add-on. But thanks to the XML method I was able to test/deactivate quickly and even at a modest add-on airport I noticed the FPS problem. In 5.2 locked to 29FPS, now with the same settings around 16FPS or worse. And that with the default aircraft. To me, however, it does not seem very fluid. Oh well...

I will now activate Hyperthreading again and see if the automatically set entries are not optimal for my system (7700k, HT still off):

[JobScheduler]
AffinityMask=15
P3DCoreAffinityMask=15
MainThreadScheduler=0
RenderThreadScheduler=2
FrameWorkerThreadScheduler=3

Affinity mask 15 on a 7700 means you have the first 2 cores + HT enabled

If you want HT disabled and use 4 (real) cores you should use 85

[Christopher Low 5,354]

I am still using P3Dv4.5HF2. I know that Kevin would probably send the bailiffs around to force me to upgrade to v5 if he could, but I am not convinced that it would be worth it with my i5 7600k/6GB GTX 980Ti powered system. I suspect that the new "core assignment" feature in v5.3 would be essentially worthless with only four cores to choose from. I personally feel that my best bet is to enjoy what I have right now, and wait to see what P3Dv6 has to offer when it arrives

[Johnny19 344]

My default p3d.cfg jobscheduler looks like this

[JobScheduler]
;AffinityMask=1048575
P3DCoreAffinityMask=1048575
MainThreadScheduler=0
RenderThreadScheduler=2
FrameWorkerThreadScheduler=4
AffinityMask=1048575
;AffinityMask=1048575

 

Any input is of course greatly appreciated. HT should be on, using i9 10900k.

 

[michaelth 76]

1 hour ago, Druce said:

So, short summary is it worth to upgrade from very stable smooth 5.2 HF2?

I am sitting in my FSL A320 in VR on the way from PAYA to CYEG at FL330 while writing. FPS is about 30 (unlocked). When I turn my head in VC view, there is no lag at all. During climb phase, I was impressed how sharp the mountains could peek through clouds. That was a big problem with EA but is solved since 5.3! 

I did make no tweaks to the Prepar3D.cfg. VolumetricClouds.cfg is controlled by ActiveSky. In outside view, there are short moments when I have the impression that small wafers begin to form, but they soon fall apart. Is that ActiveSky at work?

I won´t go back to my (stable) 5.2 HF2.

Michael

Edited December 3, 2021 by michaelth

[Cruachan 705]

26 minutes ago, Johnny19 said:

My default p3d.cfg jobscheduler looks like this

[JobScheduler]
;AffinityMask=1048575
P3DCoreAffinityMask=1048575
MainThreadScheduler=0
RenderThreadScheduler=2
FrameWorkerThreadScheduler=4
AffinityMask=1048575
;AffinityMask=1048575

AffinityMask=1048575 (decimal) = 11,11,11,11,11,11,11,11,11,11 (binary)

Doesn’t inspire me with much confidence if this is P3D’s attempt at auto optimisation for a 10 cored CPU with HT on.

Looks like we still need to do what we do best..lol

Mike

Edited December 3, 2021 by Cruachan

[kevinfirth 1,396]

2 minutes ago, Cruachan said:

AffinityMask=1048575 (decimal) = 11,11,11,11,11,11,11,11,11,11 (binary)

Doesn’t inspire me with much confidence if this is P3D’s attempt at optimising for a 10 cored CPU with HT on.

Mike

I dont think P3D is attempting any optimisation by default.  Mine defaulted to my previous 21845 AM, which was not all cores on.

[Cruachan 705]

1 hour ago, Cruachan said:

MainThreadScheduler=0
RenderThreadScheduler=2
FrameWorkerThreadScheduler=4

Do these default values refer to the assigned LPs (HT=ON) on the first 3 cores for each of the 3 ThreadScheduler categories?

Mike

Edited December 3, 2021 by Cruachan

[Sesquashtoo 2,846]

46 minutes ago, Christopher Low said:

I am still using P3Dv4.5HF2. I know that Kevin would probably send the bailiffs around to force me to upgrade to v5 if he could, but I am not convinced that it would be worth it with my i5 7600k/6GB GTX 980Ti powered system. I suspect that the new "core assignment" feature in v5.3 would be essentially worthless with only four cores to choose from. I personally feel that my best bet is to enjoy what I have right now, and wait to see what P3Dv6 has to offer when it arrives

Of course, what's always best, is what the user feels it is...the Golden Rule. Enjoy! 🙂

[kevinfirth 1,396]

12 minutes ago, Cruachan said:

Do these values refer to the assigned LPs (HT=ON) for each of the 3 Scheduler categories?

Mike

I believe so

[jlohrenz 133]

Keep in mind it is based on the MASK that you allow P3D to have access to, so if you set your masks to keep P3D off core 0, then if you have 0 for MainThreadScheduler it will actually be using core 1, if that makes sense.

If I have 4 cores, and I mask it so only 1-3 are accessible to p3d, then to the schedulers that is actually 0-2 if that helps make it any muddier. lol.  Read the burb posted a bit ago over in the prepar3d forums to understand it more.  I can see one getting easily confused here and end up causing themselves some headaches.

[vincentrouleau 71]

1 hour ago, SergioN said:

Hi, could you explain how to do it? My .cfg is showing this at the moment with HT off. My CPU is a i9 9900K

[JobScheduler]
AffinityMask=255
P3DCoreAffinityMask=255
MainThreadScheduler=0
RenderThreadScheduler=2
FrameWorkerThreadScheduler=4

Thank you

Just to make sure this new stuff is working, I put the main thread on vCpu6 and notice 100% usage on it in task manager. Also I want to make sure the main thread is not a cpu0 that can be assigned to OS stuff by windows.So I did this ( HT is on)

 

[JOBSCHEDULER]
AffinityMask=4095
P3DCoreAffinityMask=4095
MainThreadScheduler=6
RenderThreadScheduler=2
FrameWorkerThreadScheduler=4

Sometimes these subjective comments are placebo, but it feels smoother. I do not remember taxing from 18R to the gate , at night mind you, being that smooth at FT EHAM. Both that run and the run that I did on 5.2 was with 1048 textures because I run out or VRAM with FT EHAM V2 

Edited December 3, 2021 by vincentrouleau

[737freak 42]

How is VRAM usage of 5.3 compared to 5.2 and 5.1? Has it been improved?

[vincentrouleau 71]

1 minute ago, 737freak said:

How is VRAM usage of 5.3 compared to 5.2 and 5.1? Has it been improved?

I did not notice a change myself

[virtualstuff 758]

5 minutes ago, vincentrouleau said:

Just to make sure this new stuff is working, I put the main thread on vCpu6 and notice 100% usage on it in task manager. Also I want to make sure the main thread is not a cpu0 that can be assigned to OS stuff by windows.So I did this ( HT is on)

 

[JOBSCHEDULER]
AffinityMask=4095
P3DCoreAffinityMask=4095
MainThreadScheduler=6
RenderThreadScheduler=2
FrameWorkerThreadScheduler=4

Sometimes these subjective comments are placebo, but it feels smoother. I do not remember taxing from 18R to the gate , at night mind you, being that smooth at FT EHAM. Both that run and the run that I did on 5.2 was with 1048 textures because I run out or VRAM with FT EHAM V2 

Running FT-EHAMv2? with TE-NL?