Since the announcement of the Maxwell family last year and the famous VR SLI, we are still waiting to see a good use of 2 GPUs in virtual reality except on rare occasions… and that’s the case with Serious Sam VR.
There is an important group of virtual reality users whose PC barely meets the minimum requirements recommended by HTC and Oculus (at least until the arrival of asynchronous Spacewarp), and it doesn’t seem like a bad option to add a second graphics card if we already have a mid-range one with the GeForce GTX 970 or AMD’s RX-480. NVIDIA has already shown the benefits with its VR Funhouse demo, and although theoretically MultiGPU support is already available in Unreal Engine or Unity, it seems that there is still a long way to go. However, Croteam has already implemented MultiGPU mode in Serious Sam VR, which is still in early access. Croteam’s Serious Engine has always stood out for being up to date in terms of advanced graphic options (who hasn’t missed adjusting the level of detail in any game of the Serious Sam saga?), and now they prove it once again with MultiGPU support in virtual reality, both at AMD and at NVIDIA. From what has been seen, the result is quite good, managing to reduce the render time a frame from 20-25 ms to just over 10 if we use two GPUs, achieving an important increase in frames per second, which allows us to use higher levels of detail.
In traditional 2D gaming on the monitor, the use of a more powerful graphics card has always been considered superior instead of two (or more) of inferior performance, but thanks to the peculiarity of the stereoscopic 3D render of virtual reality, the time may have come to change the chip and that, as Serious Sam VR demonstrates, a second graphics card is really worth it. At the moment, there are very few experiences that really take advantage of this setup, but the trend could be reversed in the coming months if we continue to see numbers as good as those that Serious Sam has taught us. What do you guys think? Will virtual reality be the definitive push for MultiGPU technologies, which have been with us for so many years, to really be worthwhile in any scenario? Would you add a second GPU to your system if this performance improvement becomes standard?
