Pimax 8K+ VR settings i-5 12400f RTX 3070

[Richrach]

Input on a moderate system that may help others without the big pockets needed for a RTX 4090 GPU.

System:

i5-12400F (Overclocked to base 4GHz, manually set to all cores, turbo mode off)

RTX 3070 (Overclocked GPU 2115Mhz (+200MHz), VRAM overclocked to 8Ghz (+1000 MHz)

32GB DDR4  RAM

Pimax 8K+ (No base stations, 9-axis)

MSI Z690 Pro WiFi DDR4 Motherboard

Results:

Syria map, heavy mission with high load of aircraft, vehicles, and static objects

Cockpit view (A-4 Skyhawk) - 43-51FPS (Lower values only momentary.  Majority of time is 48-50FPS)

External view (F2) - 36-51FPS (HIGHLY dependent on what one is looking at.  Planes in flight upper numbers.  Helos on ground or aircraft taxiing lower range FPS)

Vehicle view (F7) - 35-53FPS (HIGHLY dependent on what is going on in the view.  Simple vehicles/static objects, no problem.  Combine moving vehicles, explosions, firing, flames, aircraft flying overhead and the frame rate hits the lower end occasionally.)

NVidia Settings:

Image scaling - Off

Anisotropic filtering - App controlled (***Make sure you have only one program doing this function.  It is easy to slip up and have several programs selected to do the same thing!***)

Antialiasing - FXAA - Off

Antialiasing mode - App controlled

CUDA- GPUs - All

Max frame rate - Off

MFAA - Off

Texture filters (in order) - Off/Allow/High performance/On

Threaded optimization - On

Triple buffering - Off

Vertical sync - Off

VR pre-rendered frames - 1

OpenComposite settings

OpenXR

OXRMR settings:

Rendering custom scale - 115% (***Make sure you have only one program doing this function.  It is easy to slip up and have several programs selected to do the same thing!***)

Motion reprojection - Disabled

Pimax settings:

72Hz

FOV - Normal

Render quality - Custom 1

Fixed foveated rendering - Balanced

Parallel Projection selected

OpenXR toolkit:

Upscaling/sharpening - CAS

Sharpening - 50%

Fixed Foveated Rendering - Off (***Make sure you have only one program doing this function.  It is easy to slip up and have several programs selected to do the same thing!***)

Turbo mode - On

Post processing - On

DCS settings (off unless otherwise listed):

Textures - Medium (cannot see more detail at high setting)

Terrain - High

Water - Med

Visibility range - Med

Shadows - Med

Res cockpit - 1024 (really cannot tell difference between this and 512)

MSAA - 4X (GREATLY reduces shimmering in headset.  I have tried all combinations of rendering levels in DSC/Pimax/OpenXR, SSAA, MSAA. For whatever reason, this seems to work best on my system.  ***Make sure you have only one program doing this function.  It is easy to slip up and have several programs selected to do the same thing!***)

Clouds - Low

Clutter/grass - 500

Forrest Vis - 50% (Syria map has WAY more trees that the Golan Heights has in reality.  I turned this down just to make it more realistic.  Higher settings do not seem to impact FPS.  Literature indicates the trees are actually still there at lower settings but not rendered to the screen)

Forrest details - 1

Scenery details - 1

Preload radius - 100

Chimney smoke - 1

Anisoptrophic filtering - 8x (***Make sure you have only one program doing this function.  It is easy to slip up and have several programs selected to do the same thing!***)

Terrain object shadows - Flat

Rain droplets - On

VR density - 1.0

Force IPD distance 40 (I flew the A-4 in real life.  Setting this to 40 gives my eyes the same perspective I had in the actual cockpit.  Your mileage may vary)

I have spent many hours working all the settings.  Much of what goes into making DCS perform at its best is FM (Freakin' Magic) and pure trial and error.  Take my inputs with a grain of salt.  I hope these give you at least a good place to start from.

Good hunting,

Richrach USN(ret)

Actual aircraft flown: TA-4J, F/A-18, F-16, B-52, SH-60B, SH-3, E-2C, EA-6B (among many others)

 

 

 

 

 

 

Edited June 25, 2023 by Richrach
Wrong words in a few places. Poor proof reading on my part.

[darkman222]

Thanks.
For the guys with the big pockets aiming at 90 fps (Hz):

Varjo Aero ( I also have a Pimax 8KX), RTX 4090, Intel 13900KS

Summary: 90 fps possible, GPU can sometimes be the bottleneck, but set to 3000x3000 pixels is "low" enough that the GPU is not getting in the way.

CPU bottleneck is still very often the case. With the 13900KS on 4ya Caucasus Server I might say like 30 % of the time youre getting in CPU frame times over 11 ms which causes stutter. Worst example 4ya Syria server going up to 16 ms. Permanent stutters.

Single player missions highly depend on how they are setup.

A sinlge F16 on Caucasus with no other aircraft in SP goes to a frame time of 4,5 ms which is a butter smooth 90fps experience. But no mission is setup like that of course.

[Richrach]

Darkman, I envy you the 4090.  By the time I can afford one of those they will have brain implants and computers will be a thing of the past.

 

With my CPU manually set to run all cores equally in power there are zero issues with CPU bottleneck.  CPU max load across the chip is 73-75%.  The GPU runs above 90% most of the time, usually at 98%.  VRAM is maxed out (8GB) and I think this is the real culprit.  How much of your VRAM is being used?  Is it near 100%?

 

I am going to hold out for a 3090 with 24GB RAM.  This is probably the real solution for DCS at this juncture for me.

Candygram!

[darkman222]

VRam is always near 100%. I also suspect this a reason for stuttering.

Regular RAM is only maxed out on MP Missions on the Syria map.

[WipeUout]

22 hours ago, Richrach said:

Fixed Foveated Rendering - Off (***Make sure you have only one program doing this function.  It is easy to slip up and have several programs selected to do the same thing!***)

I assume you are running DCS-MT?  The Pimax FFR does not work with MT, only the OXRTK FFR will work and I use the performance preset.  Did you test this?  It might be different for an 8K+ vs 8KX.

[Richrach]

WipeUout, you are exactly right.  I have FFR off in Pimax.  It was a previous set of testing I took the data from.  Everything else was correct.  Thank you for being so diligent!

Magnum!