As far as I don't it doesn't and I can't find it for OpenGL. NVIDIA exposes it via both OpenGL (IIRC NV_viewport_array2) and DX (via NVAPI). With Pascal NVIDIA has SMP (to MPA) which is adds even more. There is a reason why the 1060 beats the Fury X in the most demanding VR titles these days. Yes they both run at 90fps but 90-98% of the time the AMD hardware resorts to reprojection.
Oops, my memory failed. You're right, there's no equivalent with AMD, yet. There's equivalent for other VR things like ATW and direct mode, which I tend to forget they're driver specific features.
As far as I don't it doesn't and I can't find it for OpenGL. NVIDIA exposes it via both OpenGL (IIRC NV_viewport_array2) and DX (via NVAPI). With Pascal NVIDIA has SMP (to MPA) which is adds even more. There is a reason why the 1060 beats the Fury X in the most demanding VR titles these days. Yes they both run at 90fps but 90-98% of the time the AMD hardware resorts to reprojection.
Oops, my memory failed. You're right, there's no equivalent with AMD, yet. There's equivalent for other VR things like ATW and direct mode, which I tend to forget they're driver specific features.
Yeah direct mode is "easy" it's a simple driver hack, ATW is kinda well meh, ATW is the oculus term for reprojection which means you don't hit the frame target so you have to resort to reproject existing frames, ATW reduces the judder by a bit but it still generates pretty bad artifacts this isn't something you ever want to use unless you really have too because your hardware can't hit the true 90fps target.
Viewport multicasting and other true load reduce techniques are huge, NVIDIA had a demo with 2 VR headset running of a single GPU (I believe it was one of the newer Pascal Tesla's) and they showed that under worse case scenarios you get about the same load on the GPU as a single headset on a GPU without VPMC and multi-viewport shader pass through and under most circumstances they only were running at 30% additional graphics load because they could stream the entire 3D scene space only one, transform it once, and then do their pixel reduction screen lense transformation to reduce the amount of pixels that need to be calculated and drawn by about half.