Creating AAA titles that allow for the full range of traditional player activities is going to be quite challenging. Games like The Climb, which focus on delivering an excellent VR simulation while doing a handful of specific tasks is the kind of content we expect to see a lot of in 2016-2017.
Both NV and AMD have their own multi-GPU VR software suites (GameWorks VR and LiquidVR, respectively), and we’ve seen some VR demos running on AMD multi-GPU configurations, but we want to see shipping software and implementation before we recommend going the multi-GPU route. I wouldn’t count on multi-GPU configurations to offer huge performance gains over single cards at this stage of the game we simply don’t know enough about how well such configurations will work to have much to say about them. These are at the bottom of what Oculus recommends, but it’ll get you in the ballpark. If you can’t afford to go Fury or Fury X, I’d pick up one of AMD’s 8GB GPUs like the R9 390 or R9 390X.
The 4GB frame buffer on those two cards shouldn’t be an issue for the Oculus Rift’s resolution. If you’re an AMD fan, I’d buy a Fury or Fury X, if at all possible. If you’re wanting Team Green, I’d buy a GTX 980 or better. Oculus specifies a GTX 970 minimum target, but that card has some issues related to its 3.5GB+512MB frame buffer configuration. In that case, I’d buy as much GPU as you can afford. Even if the Oculus Rift debuts before either AMD or NV launch new GPUs, I’d still wait.
And when it comes to maintaining that 90 FPS target, or having enough headroom to spring for future VR headsets with higher-resolution displays, I’d wait. AMD and NV may be able to build wider GPUs this time out, though the fact that modern 14nm processes use a hybrid 20/14nm approach will limit this somewhat.Įven so, it’s not unreasonable to think top-end GPUs in 4-6 months will be 20-40% faster than current hardware. Both AMD and Nvidia are planning to launch new architectures in 2016, and while I can’t quote you exact specifications or performance yet, we know the general 14nm plan is to offer substantially reduced power consumption at the same performance, or a modest 20-35% uplift in performance at the same power. Lift can be calculated with the following equation: L 1 / 2 v 2 sC L. Most of this force is lift from the wings. So at MTOW, the 747 must develop about twice as much lift than closer to OEW. My advice probably won’t make AMD or Nvidia very happy, but if you’re serious about jumping to VR and you want a system that’s going to last you a while, I wouldn’t be lining up to buy a new GPU just now. The rotation speed Vr is the point at which the aircraft can pitch up and develop enough lift to climb.