Good news: I almost achieved 90 FPS with Orbx TrueEarth Netherland on Pimax 5K+

  • My computer rig: 9900K, ZOTAC RTX 2080 Ti, 36G DDR4-3600, Gigabyte Z390 Aorus Elite mainboard.

    Pitool 1.0, SteamVR 60%, 2984X2549

    Pimax 5K+


    Previous setup and test result with Aerofly FS 2 and X-plane: http://forum.pimaxvr.com/t/its…ew-update-with-9900k/8889


    Last time I ran Aerofly FS 2 with stock clock of 9900K, and the FPS is 76. The graphics is all maxed out except for shadow which is set to medium. The location is Barcelona Airport of Orbx Netherland TrueEarth, the most heavy load DLC of FS 2.


    Today I overclocked all cores of 9900K to 4.7ghz with all other settings the same as above, and the FPS is 88.9.

    Also I ran X-plane. The location is London City Airport of Orbx TrueEarth UK South, the most heavy load DLC of X-plane. Middle high graphics setting. The FPS is 29.5.


    Conclusion: Aerofly FS 2 is the most well optimized flight simulator today. With 9900K and 2080Ti, It can run at 90 FPS with the highest graphics setting in the most complicated scenery on Pimax 5K+. It's obviously making good use of all the cores of the CPU.



  • I'm pleased to see Steam VR at 60% too. If you could get to 100% you'd be using all the pixels at native res.


    I find it interesting how the CPU is a factor - we're always hearing how the CPU sits quite relaxed in FS2.


    Your previous report was Steam VR 40% - have you noticed extra clarity moving to 60% ?


    The other take away from this is that holding out for the 8Kx is perhaps not a good idea. Until we have foveated rendering no GPU will be able to cope for a good while to come with 8k pixels natively rendered.

  • I was running quite happily in a Windows Mixed Reality headset with an early i5 and GTX 970. The WMR headsets (except Samsung) have two 1440 X 1440 displays, whereas Pimax 5K+ has two 2560 X 1440 displays, less than 1.8 times the number of pixels, so why the huge increase in grunt needed? Sounds like the drivers are very sub-optimal. The AFS2 rendering engine is so efficient, I would imagine that if they can't run this, they must be pretty useless with anything else.

  • GrahamD,

    Some of the variables in VR system testing include the following:

    1) CPU - number of cores, cache size, and speed of each one (Ghz) + memory speed

    2) GPU - clock speed and number of ROPs, etc..

    3) SuperSampling value (1.0 to 2.0)

    4) VR HMD pixel dimensions ratio (actually being used at the HDMI connector) ( as you said 1.8 times for 1st gen WMR vs. Pimax 5k+ - that's quite a bit! )

    5) HMD update rate 90 Hz vs ASW (or equivalent) 45 Hz

    6) Nvidia/AMD driver version

    7) Steam VR vs. Oculus vs WMR software runtimes

    8) The application itself may handle different combinations better (headroom??) - especially as they near going slower than 90Hz (11 msec period)


    So to be truly a detailed match using a single VR application, all of these (and maybe more) have to be specified.

  • frui,

    Glad you are getting the 90fps - how much effect does using a more complicated aircraft like the q400 have on the VR fps? Also, how about the ORBX Innsbruck scenery?

    How acceptable (image quality ) is the 45fps (ASW equivalent half frame generation) mode look on the Pimax 5k+ ?

  • frui,

    Glad you are getting the 90fps - how much effect does using a more complicated aircraft like the q400 have on the VR fps? Also, how about the ORBX Innsbruck scenery?

    How acceptable (image quality ) is the 45fps (ASW equivalent half frame generation) mode look on the Pimax 5k+ ?

    I am not into q400 (too complicated for me) but I will give it a try. Orbx Insbruck should be no problem as it's less demanding than Amsterdam TrueEarth Netherland. Pimax 5K+ does not have ASW (yet), but at 45fps it's quite playable. I suspect they have implemented some kind of partial ASW of their own, but they will push forward something called "brainwarp" in the future.

  • I take your point entirely, but with roughly 3 X the raw processing power, the RTX 2080 Ti should cope quite easily with a 1.8 X increase in pixel count, though I admit that with high supersampling and dense scenery the 970 may have struggle a bit). The problem may be that when you approach 180 degrees, you can no longer project onto a flat plane, so a major shift in rendering technique may be required. I wouldn't be surprised things steadily improve with time,

  • GrahamD

    Most benchmarking of the 2080ti over the 1080ti are showing a modest 30% or so improvement in performance (everything else the same).The 3X stuff has to do with ray tracing.


    Example:

    https://www.techradar.com/news…-ti-vs-nvidia-gtx-1080-ti


    frui

    I think a lot of folks were getting 5.0 Ghz for all cores but maybe they are using special cooling. 4.7 is very fine though. The major point that you are showing is simply that we need 6 or more 5.0 Ghz or so cores plus a 2080ti + Pimax 5K+ plus controllers and sensors - WOW!!

  • GrahamD

    Most benchmarking of the 2080ti over the 1080ti are showing a modest 30% or so improvement in performance (everything else the same).The 3X stuff has to do with ray tracing.

    I was referring to performance compared with a GTX 970, which ran adequately. At the moment, it seems that the Pimax needs an absolute top end graphics card and CPU to have any hope of running fluidly. This would suggest that a wide-angle VR system is going to require a major investment, even for something like AFS2 with a very efficient rendering engine. From pixel numbers alone I would have expected at GTX 1080 or 1070 ti with an i5 to be adequate. If/when occulus and Vive have wide angle systems, things may change.


  • I have achieved 5.0ghz on all 8 cores with a cheap watercooling, but the gain of fps is very slight compared with 4.7ghz all cores, around 2 fps, at the cost of higher voltage. So I decide it's not worth the extra power consumtion and heat. I 4.7ghz all core at 1.27v is some kind of sweet spot for Aerofly FS 2, but I will explore more.