Some P3D Performance Testing

[TravelRunner404 596]

Hey Everyone,

I have a new install of P3D and thought I would do some testing with FRAPS to see how sliders and certain add-ons affect performance in a clean P3D install.  I created a recorded flight from KAPA out over central Denver.  Went to about 10k feet then back down to 7k.  In Denver this only about 2k-5k feet above the ground.  Test flight was about 4 minutes long.  I am not trying to make any hard claims with this but I think it could be a good baseline to understand how sliders affect your overall performance in P3D.  This had nothing to do with FPS as a whole, I am well aware of the 30/30.  I am not complaining about performance.  Simply trying to establish a baseline then see how performance changes with settings.  Hopefully, you will find it useful.

My system: 8700k OC @ 4.8ghz, Geforce GTX 1080ti, 32mb RAM, 1TB SSD, 4k monitor

Impressions:

-LOD Radius is about the only thing that affected performance in Terrain sliders.  Everything else on the highest or lowest setting was within the margin of error.

-AGN Draw distance and AGN Building/Tree Density were the biggest impacts in Scenery Objects.  Moving any slider to normal in scenery objects affected performance within margin of error but buildings and draw distance was the biggest.

-Shadow Quality and Distance from Ultra to Medium had surprisingly low impact and my 1080ti was at 99-100% on baseline test.  It was right at the margin of error, both together might lend 2% improvement.

-SSAA is a killer but that was known.  8xMSAA vs 4xMSAA was negligible and right at the margin of error.

-Unchecking High Res Terrain textures made performance worse.  This was repeatable and outside margin of error.

-Adding reflections (Not Dynamic) was a huge performance hit.  Baseline I had the standard clouds and vehicle selected.  Turning off clouds and only having vehicle selected added no performance boost.  

-Road Traffic in my scenario was about 3.2% drop in performance per 10% Road Traffic added.

-Orbx Vector did absolutely nothing to performance.  I know there is information out there that it seems too but I repeated this test multiple times.  Added it to NA LC and even then the change was within the margin of error.  I don't think Vector affects performance but more testing maybe.

-The Central Rockies region performs better than NA LC, Vector and Base.  Remember vector data is included in Orbx Regions.  Remember all products were installed stand alone

-Special effects low vs medium did nothing in this scenario.

Images of test results:

https://ibb.co/LdmX7vR

https://ibb.co/y4D0mwP

 

[itsjase 52]

11 hours ago, TravelRunner404 said:

Shadow Quality and Distance from Ultra to Medium had surprisingly low impact and my 1080ti was at 99-100% on baseline test.  It was right at the margin of error, both together might lend 2% improvement.

This is only the case during the day. Try at dusk or dawn when there are long cloud and scenery shadows and there will be a much bigger difference.

[AnkH 2,232]

You tested reflections from the menue with check boxes, right? What is about this dynamic reflection slider? I currently have it on medium, one reason is the often exaggerated reflections, the other one because I think it is hard on performance. Can you confirm this?

[jcomm 5,882]

I'm just waiting for 4.5 to get released, and then will re-install a P3D 4 using a monthly license.

Usually I unselect reflections and all sorts of traffic, so, performance isn't really an issue...

Only "heavy" stuff I usually have in my temporary P3D installs is ORBX FTX Global and Madeira scenery ( Aerosoft ). Sometimes a free EDDF too... Then PMDG 777 and the FSLabs airbuses...

It all depends on what you're up to, given the limitations imposed by your hardware. I don't want to invest in my 2011 bought machine, i5 2500 and GTX 960, so, I can't ask for much.... 

[itsjase 52]

3 hours ago, AnkH said:

You tested reflections from the menue with check boxes, right? What is about this dynamic reflection slider? I currently have it on medium, one reason is the often exaggerated reflections, the other one because I think it is hard on performance. Can you confirm this?

The dynamic reflections just determines what is reflected on aircraft with dynamic reflections enabled. Mostly PBR planes.

Low: basic world features (ex: terrain, sky, water)
Medium: taxiway and runway features)
High: autogen buildings and air features (ex: stars, clouds)
Ultra: details (ex: particles, vegetation)

[TravelRunner404 596]

If anyone is curious I did the baseline in a fresh P3D v4.5 install and it was 3 FPS less. 

[fluffyflops 1,843]

On 4/7/2019 at 7:47 PM, TravelRunner404 said:

Hey Everyone,

I have a new install of P3D and thought I would do some testing with FRAPS to see how sliders and certain add-ons affect performance in a clean P3D install.  I created a recorded flight from KAPA out over central Denver.  Went to about 10k feet then back down to 7k.  In Denver this only about 2k-5k feet above the ground.  Test flight was about 4 minutes long.  I am not trying to make any hard claims with this but I think it could be a good baseline to understand how sliders affect your overall performance in P3D.  This had nothing to do with FPS as a whole, I am well aware of the 30/30.  I am not complaining about performance.  Simply trying to establish a baseline then see how performance changes with settings.  Hopefully, you will find it useful.

My system: 8700k OC @ 4.8ghz, Geforce GTX 1080ti, 32mb RAM, 1TB SSD, 4k monitor

Impressions:

-LOD Radius is about the only thing that affected performance in Terrain sliders.  Everything else on the highest or lowest setting was within the margin of error.

-AGN Draw distance and AGN Building/Tree Density were the biggest impacts in Scenery Objects.  Moving any slider to normal in scenery objects affected performance within margin of error but buildings and draw distance was the biggest.

-Shadow Quality and Distance from Ultra to Medium had surprisingly low impact and my 1080ti was at 99-100% on baseline test.  It was right at the margin of error, both together might lend 2% improvement.

-SSAA is a killer but that was known.  8xMSAA vs 4xMSAA was negligible and right at the margin of error.

-Unchecking High Res Terrain textures made performance worse.  This was repeatable and outside margin of error.

-Adding reflections (Not Dynamic) was a huge performance hit.  Baseline I had the standard clouds and vehicle selected.  Turning off clouds and only having vehicle selected added no performance boost.  

-Road Traffic in my scenario was about 3.2% drop in performance per 10% Road Traffic added.

-Orbx Vector did absolutely nothing to performance.  I know there is information out there that it seems too but I repeated this test multiple times.  Added it to NA LC and even then the change was within the margin of error.  I don't think Vector affects performance but more testing maybe.

-The Central Rockies region performs better than NA LC, Vector and Base.  Remember vector data is included in Orbx Regions.  Remember all products were installed stand alone

-Special effects low vs medium did nothing in this scenario.

Images of test results:

https://ibb.co/LdmX7vR

https://ibb.co/y4D0mwP

 

thanks are you using an affinity mask ?