nVidia's "Single Pass Stereo" VR Performance in iRacing

Here’s an interesting investigation of iRacing VR performance on nVidia cards:

TL;DR: “…there is a near-zero difference between SPS On or Off.”


Well, that puts the final nail in the coffin for that particular VRWorks feature then, considering the dissertation the author of this article references, had already not only suggested as much, but also pointed out the multiple manners in which it, due to its presuming symmetrical frusta, tends to lead to incorrect rendering (especially for the right eye). .

I don’t believe every title that renders e.g. ambient occlusion wildly differently between the eyes, are using SPS – developers are probably quite able to mistakenly reuse a single frustum for the view-dependent fragment shader runs for both eyes, without NVidias help, but it’s a start – could maybe bring some much needed attention to the matter. :7


All i do is irace in vr. the 10-15 percent performance you get with SPS can be the difference to getting solid or mostly 90 and the fps fluctuating all over the place.

Also, given architectural differences between sps on the 20 series and 10 series cards … there are additional settings only available in INI files to change hoe the eyes are rendered. There are 3 profiles, 1080 users are suppose to figure out which one fixes the image for them.

But, as he noted iracing is a mostly singlethreaded game, so SPS isnt goimng to get you to constant 90… because your CPU is you bottleneck. If you can get smooth 90… SPS will allow you to turn on a couple more graphical features.

There are multiple spreadsheet out there where people have gone through the settings one by one to show the performance impact per settings.

Also it looks like he might have ovr toolkit running which also taints the numbers.

The only data he posted from the runs that matter are the Frametime numbers(specifically the worst frame times) . So, when i configure iracing I try to grab a demo from the most cpu heavy recording i have… then try to find the graphics settings to ensure i dont drop any frames. In my cause the headset is set to 110 hz.

Then I leave those settings in play until the next update.

If you’ve never optimized your system to get max fps always. You’ll never see the real benefit of the extra 10%.


Absolute FPS isn’t always a great measure of performance in VR. SPS helps considerably with frame time consistency. Always surprises me when I see reviews that use absolute FPS as the one and only measure of quality and comfort.

A lot of reviews don’t even take into account when comparing video cards frames that are generated via ASW and those that are fully rendered. Heck I see reviews where frames are regularly dropped but the absolute FPS is still high and they say there are no differences when really there are significant differences.