Thanks Ray yes a complete reinstall is probably what I am looking at. I am not convinced that an addon is to blame as there would have been others complaining. I have been testing with an inflight scenario that involves the PMDG 737-800 doing a  an anti clockwise circuit around YSCB (Orbx) departing and arriving on rwy 35.  This airport and associated scenery is unremarkable but since the earliest days has been a FPS hog. With P3D4.5 I would have mid 20s at takeoff then a steady increase to 40 plus once well to the south of the airport.  On approach frames would drop to the low 20s until over the fence and back to mid 20s on the runway,  When I initially upgraded to P3D5 I got about a 15% increase on what I could get in P3D4.5 using equivalent settings that included scenery complexity at very dense.  However now after takeoff FPS is all over the shop from low 30s to single digits -  this creates annoying stutters and it never increases to 40 once south of the airport. As previously noted I have to wind scenery complexity back to sparse to get acceptable performance and the problem with that is many payware airports can look very ordinary. One setting that isn't the same as what I previously had is EA and volumetric clouds but that is only because if I turn off volumetric clouds performance is even worse!  Go figure. 

Bruce

Bruce, I have YSCB but haven’t flown there in a while. I also have YBBN v2 and in the PMDG737 was having horrendous stutters. There weren’t there with a small GA. But Brisbane isn’t a complex airport.

It turns out the Dynamic Reflections setting was the reason. Turned it from High to Medium and it became totally smooth again. I think that’s the name of the setting. Check you don’t have it above Medium.

There is surely something basically wrong with your setup.. I only have a 1080p display with a GTX1080 card but with scenery complexity at one below maximum , at your test YSSY 16R, with PMDG 747, I get between 28 and 30 fps with no trouble. (All other settings are to the right of middle, full right would tax my 8Gb of vram).

On 2/3/2022 at 8:26 AM, brucewtb said:

My upgrade path to 5.3 initially was to unistall client, content and scenery then reinstall followed by deleting Prepar3D.cfg

Deleting Prepar3d.cfg means setting all your preferences again. This makes me think they may be different to before you updated unless you took screenshots of each setting page. Did you?

2 hours ago, IanHarrison said:

There is surely something basically wrong with your setup.. I only have a 1080p display with a GTX1080 card but with scenery complexity at one below maximum , at your test YSSY 16R, with PMDG 747, I get between 28 and 30 fps with no trouble. (All other settings are to the right of middle, full right would tax my 8Gb of vram).

Hi Ian is that with Fly Tampa YSSY V2 and Orbx Australia V2?  Just to be clear I can max out all the GPU dependent settings as you would expect with a 3080ti but Scenery Complexity is CPU dependent.  Overall any GPU from a 1080 upwards can run P3D it doesn't make much difference but CPU is another story although my 9900KF should be able to do better than what I am seeing and has done so with previous P3D versions.

Ray I have had Dynamic reflections disabled.

Bruce

50 minutes ago, brucewtb said:

Ray I have had Dynamic reflections disabled.

In that case a new install is my only suggestion. To a new location too. Ideally D:\P3Dv5 or something very similar.

2 hours ago, brucewtb said:

Hi Ian is that with Fly Tampa YSSY V2 and Orbx Australia V2?  Just to be clear I can max out all the GPU dependent settings as you would expect with a 3080ti but Scenery Complexity is CPU dependent.  Overall any GPU from a 1080 upwards can run P3D it doesn't make much difference but CPU is another story although my 9900KF should be able to do better than what I am seeing and has done so with previous P3D versions.

Ray I have had Dynamic reflections disabled.

Bruce

Yes. Fly Tampa v2 and Orbx v2. But I have turned the YSSY options down a bit by turning off 3d people.

That is what I mean, you have a 9900KF running at 5Ghz, mine is overclocked at 4.3. Something has to be wrong.🤔

I think Ray may be right, a full removal and installation would clear any doubt.

Edited February 6, 20224 yr by IanHarrison

9 hours ago, IanHarrison said:

Yes. Fly Tampa v2 and Orbx v2. But I have turned the YSSY options down a bit by turning off 3d people.

That is what I mean, you have a 9900KF running at 5Ghz, mine is overclocked at 4.3. Something has to be wrong.🤔

I think Ray may be right, a full removal and installation would clear any doubt.

I don't have my 9900KF overclocked because I had stability issues with an OC of 5.1 ghz and MS2020 but with turbo boost the CPU runs between 4.8 and 5.0 ghz which is close enough - so I have stuck with that rather than trying to debug the OC.  Now I am running at 4K which means complex airports are a challenge but quite achievable even with scenery complexity at very dense - or was so prior to the P3D5.3 update and still is with P3D4.5 and MS2020 (although with that I get much more benefit from the 3080ti).  Now I have had issues getting a matched checksum with the  P3D 5.3 full download zip but the one I used for my latest install does have a matched checksum.

Bruce

3 hours ago, brucewtb said:

 Now I have had issues getting a matched checksum with the  P3D 5.3 full download zip but the one I used for my latest install does have a matched checksum.

Bruce

One more thing you could try. I had an issue a little while ago where we had a power cut and I was in the middle of a flight and couldn't get it closed down in time before the UPS ran out.

After that, I kept having unexplained failures such as P3d just crashing after start or not even starting. I did a repair on P3D but still had problems. Then I repaired Windows and all was and remains stable again.

I am not saying that your problem is the same but have you tried the Windows repair procedure (DISM and sfc/ scannow) just to make sure all is in order on that side?

Well I have done a reinstall by uninstalling the four components in the Control panel and running a freshly downloaded full install  file with a matched checksum, Now I can set up the 747-800 scenario at 16R YSSY used previously and get FPS similar to or better than with P3D5.2 or 4.5.  So success of a sort.  However I am still left with degraded in flight performance with the PMDG 737NGXu compared to P3D4.5 using my YSCB scenario.  I am going to have to test this more widely with scenarios from other locations.  What I have also noticed is that my 3080ti is only running at 30% whereas the CPU remains maxed out at 100% on at least one core - very different to the way MS2020 works.  One other thing is that it seems to me P3D5.3 has been optimised for EA and volumetric clouds which means you get lower performance with EA off and 2D clouds - makes comparisons with 4.5 more problematic.  I think I am just about done with this I am not sure whether I am getting worse performance than previously with the new version 5.3HF3 but it is different.  I don't care for the very hazy look you get with EA although through the gloom terrain colours are more realistic than P3D4.5 something akin to what you see in MS2020.  But it seems you have to leave EA and volumetric clouds on or incur a FPS penalty with them off

Bruce

 

 

1 hour ago, brucewtb said:

Now I can set up the 747-800 scenario at 16R YSSY used previously and get FPS similar to or better than with P3D5.2 or 4.5.

Excellent!

1 hour ago, brucewtb said:

What I have also noticed is that my 3080ti is only running at 30% whereas the CPU remains maxed out at 100% on at least one core - very different to the way MS2020 works.

My 1080Ti is the same. Open Prepar3D.cfg and find the JobScheduler section. Paste the contents here together with your CPU description. LM have changed the way P3D runs depending on how many cores your CPU has. Do you have Hyperthreading enabled in the BIOS.

I have EA off. It’s better that way. There’s certainly no penalty for that.

19 hours ago, Ray Proudfoot said:

Excellent!

My 1080Ti is the same. Open Prepar3D.cfg and find the JobScheduler section. Paste the contents here together with your CPU description. LM have changed the way P3D runs depending on how many cores your CPU has. Do you have Hyperthreading enabled in the BIOS.

I have EA off. It’s better that way. There’s certainly no penalty for that.

I currently have HT on but sceptical about any benefit from this.  As for EA with my system there is certainly at least a 5fps penalty for having it off if 2D clouds are used but no difference if the clear weather preset is used (ie no clouds).  I am curious as to whether others see this penalty for using 2D clouds instead of volumetric clouds as it is quite counter intuitive.. I use the Job Scheduler settings suggested by Steve earlier in this thread. 

[JobScheduler]
AffinityMask=65493
P3DCoreAffinityMask=65493
MainThreadScheduler=0
RenderThreadScheduler=1
FrameWorkerThreadScheduler=2

I have followed the various "Job Scheduler" threads on this forum with interest but I have to confess they go over my 76 year old head.

Now if I do a test flight with the NGXu taking from FlyTampa's YSSY rwy 16 to the south and then doing a 180 over the city and heading south west towards Canberra my FPS in P3D5.5HF2  is in the low 30s on the runway and climbs to the mid to high 30s at altitude but can vary a lot and drop momentarily into the 20s and teens.  Whereas with P3D4.5 FPS  is in the mid 20s on the runway but climbs to the high 40's at altitude and stays there.  I use the Clear weather preset in both.  I think I prefer P3D5.3HF2 with EA on despite the very excessive haze but that has more to do with my 49" Sony TV's colour settings (something called colour vivid) which are optimized for photo editing but in P3D5 results in a highly saturated display with EA off.  This saturated effect is somewhat less of an issue in P3D4.5 and MS2020.

So has performance got worse with P3D5.3HF2 compared to 4.5?  I guess the answer is "possibly" and "yes and no".  Does it matter?  Well provided I can keep FPS above 30 then no but in some situations this is not the case in P3D5.3HF2 compared to P3D4.5.

Bruce

 

 

Bruce, your report is so subjective it’s hard to fathom out if there’s been any improvement. Go with what looks best for you or what gives you the best frame rates. Only you can decide that.

If Steve has given you his recommended settings for JobScheduler then you can do no more. Just adjust FlyTampa settings and turn 3D people OFF as it has a big impact.

One thing though. In your first post you said you have a 30Hz display and run with VSync on and frames unlimited. That limits your frame rate to 30 so what have you changed to have fps in the “mid to high 30s”?

@brucewtb,

P3D v5 has roughly 4/3 the scenery load compared to P3D v4 but the rendering performance has improved a bit. So maybe you will see better performance in some cases and not in others depending on the settings chosen and add-ons.

Affinity Masks and HyperThreading have been knocked by all sorts of folk and this has caused confusion on the subject for years. Put simply, with HT enabled you have to be careful not to lose fps, but won't gain you fps (possibly one or two). You can gain a bit (20% - 30%) in scenery loading (fill-in) and SimConnects performance with the extra HT threads reading files and networking which is highly threaded. The affinity settings you chose (65493,65493,0,1,2) will be as good as you can get with 8 cores/16LPs. With HT disabled you would do 255,255,0,1,2 - this would reduce file and network performance slightly (SimConnects run on networking even if there are no other PCs).

Edited February 9, 20224 yr by Ray Proudfoot
Edited to remove long quoted post

16 hours ago, Ray Proudfoot said:

Bruce, your report is so subjective it’s hard to fathom out if there’s been any improvement.

One thing though. In your first post you said you have a 30Hz display and run with VSync on and frames unlimited. That limits your frame rate to 30 so what have you changed to have fps in the “mid to high 30s”?

Well in these tests I have been relying on what the FPS counter was telling me which I would have thought was quit objective rather than a subjective what looks best (but I agree at the end of the day "what looks best" is the test that matters). The frames in the mid to high 30s refers to the test situation which I run with vsync off - my usual practice is to have vsync on so I get fps of 30 (hopefully) which is the refresh my Sony TV is set to.

 

13 hours ago, SteveW said:

@brucewtb,

P3D v5 has roughly 4/3 the scenery load compared to P3D v4 but the rendering performance has improved a bit. So maybe you will see better performance in some cases and not in others depending on the settings chosen and add-ons.

I guess that is the case. 

Bruce