Microsoft announced a year ago DirectX raytracing and promised hardware-accelerated raytracing graphics for PC games. In August, Nvidia announced the new RTX 2080 and 2080Ti video cards with the company's new Turing RTX processors. In addition to the regular graphics processing hardware, these new chips contained two additional sets of additional cores, one set for the execution of machine learning algorithms and the other for the calculation of raytracing graphics. These cards were the first and only cards currently supporting DirectX Raytracing (DXR).
This will change in April as Nvidia has announced that 1
The performance of these cards will not match that of the RTX chips. RTX chips use both their raytracing cores and their machine learning cores for DXR graphics. To achieve a suitable level of performance, raytracing simulates relatively few beams of light and uses machine learning antialiasing to filter out the ray tracing images. Without the dedicated hardware, DXR will use 32-bit integer operations on the GTX chips on the CUDA cores already used for compute and shader workloads.
Nvidia says that Turing and Pascal cards take two to three times longer to render than any Turing RTX card. This difference is particularly evident in Pascal cards. In Turing, the 32-bit integer workload used for raytracing can run concurrently with the 32-bit floating-point workload that is used for other graphical tasks. This is not the case with Pascal, where workloads have to be executed one after the other.
Due to this weaker performance, Nvidia recommends that developers only use simpler Raytracing effects on the older chips. For the RTX parts, the raytracing performance can be good enough to allow for global illumination – a form of ray tracing that provides indirect illumination through reflections in addition to the usual direct illumination from light sources. For the GTX parts, however, Nvidia recommends the use of simpler tasks such as material-specific reflections.
This raytracing can be done with these older chips is not a big surprise. During the development of DXR, Microsoft used a ray tracing algorithm based on a computer shader, so dedicated hardware is not required. However, the significant performance difference shows that the dedicated hardware will be important, at least for the moment.
Existing games using DXR should automatically begin raytracing on the 10 and 16 cards as soon as the drivers are updated without requiring game updates. However, given the performance difference, we would imagine that developers want to adjust their raytracing to the older hardware. Raytracing puts considerable strain on the RTX chips, and it is simply not practical to use the same quality of raytracing quality on the older cards.