At its FMX graphics conference in Germany, Intel just many a surprise announcement. Although details about Intel’s Xe discrete GPUs are still scarce, the company just went ahead and said that its upcoming GPUs will support hardware level ray tracing, albeit just the Data Center parts in the beginning.
This means that Intel’s Xe GPUs will directly compete with NVIDIA’s Ampere or whatever their next-gen architecture is called. At the moment, NVIDIA’s Turing lineup is the only GPU family in the market to support ray tracing on a hardware level via the RTCores. Although now the GTX Turing and Pascal parts also allow ray tracing thanks to DXR’s software fallback solution, it is exponentially slower compared the RTX cards.
Intel might have only promised ray tracing for the Data Center space, but it’ll only be a matter of time before the technology finds its way into the consumer products. However, just the fact that Intel’s enterprise Xe GPUs will support hardware level ray tracing means that the entertainment industry might shift from using CPUs to GPUs for rendering animated movies and the like. The reason is that it’ll be like 10 times faster while costing pretty much the same (probably). Like CPU based encoding, the rendering is also more precise but now GPUs have progressed to a level where the difference is almost unnoticeable. NVIDIA’s recent NVENC addition to the GTX Turing graphics parts for streamers is clear-cut evidence that soon GPUs will become a lot more widespread compared to CPUs. The official announcement from FMX, Germany confirming hardware-level ray tracing for Intel Xe GPUs is as follows:
You can read the whole press release here.