NVIDIA’s lineup of RTX graphics cards appears complete, for the moment. With the release of the RTX Titan at the annual Neutral Information Processing Systems (NeurIPS) Conference held December 2-8, 2018 in Montreal, Canada, NVIDIA’s new Turing architecture is well represented across the high-end gaming, enthusiast, and high-performance computing (HPC) markets. RTX also further cements NVIDIA’s gradual shift in focus towards specific HPC tasks that form the foundation of some of today’s buzziest technologies, such as autonomous vehicles and artificial intelligence. However, the Titan’s release has done little to assuage concerns that bleeding edge components in Turing, namely RT Cores and Tensor Cores, do not translate to significant gains in graphics quality for VR, or midrange mainstream gaming.
RTX In Context
Compared to its ancestor, the Titan Volta (which also employed Tensor Cores, as a similarly professional-oriented HPC card), the RTX Titan positions NVIDIA hardware even further away from mass-market gaming applications. This is supported in part by climbing price points and power demands; the entry-level 2070 card receives a price bump of over $100 compared to the 1070, and even the recommended 650w power supplies recommended by NVIDIA struggle in high-end VR applications. In addition, media outlets have noted the relatively underwhelming real-life performance of ray-tracing and Deep Learning Supersampling (DLSS) features in contemporary AAA titles such as Shadow of the Tomb Raider. To its credit, NVIDIA has been vocal about the onus of ray tracing implementation resting largely with developers, and that investments in Turing architecture cards are “future-proofing” systems for when developers catch up to the hardware.
However, these developments have yet to manifest meaningful improvements in VR rendering; while the company enjoys a renewal of excitement after a recent patch brought Battlefield V to an “utterly playable” 63 frames per second (FPS), VR languishes in its attempt to deliver a consistent 90 FPS at similar resolutions. For the moment, then, it seems that RTX features hold little relevance in VR, where priorities and considerations in quality differ greatly from traditional gaming, and a “nice-to-have” becomes a critical function preventing discomfort in its users.
Greenlight Insights’ predictions surrounding the rise of standalone VR as a crucial milestone for the technology raise a pertinent question for VR industry stakeholders: in an increasingly dynamic market of midrange and last-generation cards, in which PC gaming is becoming less of a priority for GPU providers compared with servers and high-powered computing, is pushing the performance envelope ever higher a sustainable goal? Or should OEMs be working on solving different problems, such as friction, in the service of driving adoption outside of enthusiasts circles?