@Shrike
Ohh sorry, my fault,
I mistook you it was someone else’s DCS post. sorry again
2 Likes
Yeay, my fault … sorry
1 Like
Go away to troubleshoot and everything blows up 
@mirage335 I didn’t fly the hornet at night to test, sorry. I did notice that the SU25 instrument panel was tinged with a purple colour when outside the foveated circle.
1 Like
Ah, same thing, thank you for testing. Any chance that goes away with wider FOV settings? Hoping DFR chases this effect far away from the center of vision, where it can’t be seen too much.
no nothing to do with this thread, I saw the other quorra section post reporting that the eye tracking module was running at 30hz due to the USB hub in the headsets all being USB 2.0 instead of 3 and cross posted it to reddit as it was in a “private” section of the forum and I figured people ordering headsets or eyetrackers deserved to know the real specs vs. what Pimax are listing on their website
this isn’t the first time that Pimax have listed up specs wrongly and then people find out after the fact
I didn’t mention DCS is my reddit post as the opening post of the thread I looked at didn’t mention DCS either - the fact that pimax list the specs of the eye tracking as 120hz but it can only run at 30hz with all Pimax headsets is not directly related to DCS
the fact that all Pimax headsets only have a USB2 hub also affects people’s ability to run, say, eyetracking and hand tracking at the same time, or any other “modules”
1 Like
We now know that also the Vive only has USB 2.0 ports. usb 3.0 is used only for internal functions.
This report tool say, there are 2 USB 2.0 Hubs, if I remember right.
If this are correct, both 2.0 Hubs are connected to 3.0 with full 2.0 Bandwidth.
there is only one port on the 8KX
this topic is about the 5k +
okay yes, I’ll remember to keep info hidden because it offends you
1 Like
The thing is, foveated rendering is intended to help games/apps which are GPU-bound, yet DCS is CPU-bound.
I wouldn’t have expected it to improve the framerate of DCS. I wouldn’t have expected it to decrease the framerate either, but it sounds like foveated rendering places an additional load on the CPU. Trading CPU cycles for a GPU performance increase makes sense most of the time, but not when the game/app is CPU-bound.
2 Likes
Eye tracking would work best if it were a hardware chip in the hmd doing the processing it seems
1 Like
True, but that would raise the price of the module. Probably the larger issue is that sending the video back to the PC increases the data on the USB channel. That could (possibly) create issues for the controllers or hand-tracker.
Going back to the OG, what can be done to improve the DCS VR experience? It sounds like first and foremost, we need a CPU with a major improvement to the individual thread performance. Is it limited by computation or memory bandwidth? If it’s the latter, then there’s hope, since CPU caches are getting larger and main memory will soon be on the same interposer as the CPU.
DCS World, under many use cases, is nearly equally CPU core and GPU core limited. Pushing these clocks up every just 1% makes a huge difference in visual quality and reliability.
Strongly not CPU memory or GPU memory limited - pushing memory clocks up on both sides makes almost no difference at all.
DCS World not only runs just about everything through a single thread, it runs some aspects of its graphics postprocessing pipeline through that same thread. That is why pushing up high resolutions hits the CPU bottleneck, and why trading off the pixel density (built-in supersampling), helps shift that to other threads.
If they would just break out that one part of the process, we would be fine on the CPU side.
3 Likes
Hopefully Wags and Nick manage to go multi-threaded next year - it would be about time (20 years?)…
2 Likes
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.