In Pimax Refresh Rate Test 10/30 we can see that the headset is running as default in 4096x2557 pixels/per eye.
We can also see that is in 80hz which equals to 1.741 MB/Sec.
So i wanted to know if i could run Project Cars 2 (which is the main reason i want the Pimax 8K) on a
GTX 1080 TI OC, 16GB RAM, i7-7700
So i tested with my Vive, SS set to 3.8, this equals to 1.730 MB/Sec, about the same as Pimax default.
All settings were either low or off and i could still not run the game without ALOT of dropped frames.
This was also without cars and without any weather effects.
Am i thinking like a horse or should i be concerned?
yes. but there are two ways to render that resolution, your graphics card can render 4096x 2657 which requires x work(which is what you did), or it can render 1440p and upscale to feed in that resolution. which requires much less work.
edit, i think a compounding problem with pimax is communication. even when trying to communicate they will drop a video like that without explaining what we are seeing, and the implications etc. I wish they were better at laying this stuff out clearly.
You are basically thinking right. The test you did is actually the one of possible ways to get an idea how the Pimax8K will tax your machine. Assuming you were around the same refresh rate.
@destraudo The rendering resolution shown in the picture is the actual size at which the scene is rendered by gfx card. So this is the “real deal” the card has to handle. Then this image is transformed into 2560*1440 resolution by using pre-lens warp transformation and as such is sent to the HMD. Then HMD upscales the image to 4K res of the panel again.
I would take the info from steamvr with agrain of salt. As we all know, pimax uses their own renderer which does not relate in any way to these steamvr render settings.
The weird resolution alone (4096x2657) per eye with a supposed 1x resolution ss multi does not fit.
If you think there is a 1.85x multi, because 2657/1440=1.85, okay, but 4096/2560=1.6x… does not fit…
The resolution it renders should be a multi of 2560x1440 in both directions…
Yes, it is 4096*2657 per eye as noted in the screenshot you posted. But beware that both eyes are rendered from the same scene geometry with only different view port. So I would expect that the complexity to render those two frames (for the left and right eye) is not so demanding as rendering two completely independent scenes of the same size. But I have no internal knowledge about how SteamVR works, so take it only as my personal assumption.
The supersampling displayed in the slider control is different from the one used for pre-lens warp transformation. They both add in the final image.
In the picture, the “user SS” is set to 1.0, while the “SteamVR SS” is set to the values you calculated. Those values (I speculate) are derived internally by SteamVR runtime from the lens geometry the Pimax layer is reporting to SteamVR.
AFAIK they don’t. There were some videos published by Pimax from PC and Elite, but both suffer from the fact that they were shot on smartphone without a stativ and only were showing some static part of the game (e.g. hangar, or pit-stop). Also it is difficult to really evaluate 90 FPS in the video which is shot at 60 FPS. I would rather put my faith into the reviews of the users who had direct experience on some of the shows.
I think you are lucky they are showing anything. What other headset makers are willing to put their neck on the line like this? does Oculus update their development process every step of the way? Does Vive or any of the MR headset makers? no!! Just get off their backs and liet them solve the problem without being so friggin nit picky.
but the whole thing this originated from with the questioning of the pimax guy and the discussion of actual frame rates came from a session in which he got to see onward being tried and got to play project cars 2 with it. edit to say they also have video up of asetto being played, car going around corner etc.