Notebookcheck Logo

AMD sneaks additional details on the expected RX 6000-series ray tracing performance in the RDNA 2 overview

The Navi 21 compute units now include Ray Accelerators. (Image Source: AMD)
The Navi 21 compute units now include Ray Accelerators. (Image Source: AMD)
Even though AMD did not show any ray tracing game benchmarks, it at least mentions a few metrics in a footnote from the RDNA 2 architecture overview. Team Red tested an unspecified RX 6000 model with the DXR SDK more than 2 months ago and the performance seems to be a bit higher than the RTX 2080 Ti, but still lagging by 33% when compared to the RTX 3080.

We were really hoping to see some ray tracing performance metrics together with the rest of the awesome graphs shown yesterday by Team Red during the RX 6000 GPU presentation, but, unfortunately, that did not happen. However, AMD chose to sneak some performance info in the footnotes of the official RDNA 2 architecture page and it looks like the previous rumors that claimed the Navi 21 RT performance was on par with the RTX 2080 Ti were not entirely right.

First off, the RDNA 2 architecture page states that the high-performance ray tracing acceleration architecture new to the Navi 21 compute units is called Ray Accelerator. These accelerators are said to provide “an order of magnitude increase in ray intersection performance compared to a software implementation,” a statement that is accompanied by a footnote detailing the process calculating this order of magnitude. The footnote specifies that those measurements were calculated by AMD engineering labs on 8/17/2020 for an AMD RDNA 2 based graphics card. They are not mentioning the exact model, but we can presume it is one of the high-end cards, and looking at the date, AMD surely has managed to further optimize performance in the meantime, at least by a few percent.

The footnote goes on to say that the measurements were recorded “using the Procedural Geometry sample application from Microsoft’s DXR SDK, [and] the AMD RDNA 2 based graphics card gets up to 13.8x speedup (471 FPS) using HW based ray tracing vs using the Software DXR fallback layer (34 FPS) at the same clocks. Performance may vary.” Now, reddit user ET30 tested his ASUS RTX 3080 TUF card in the DXR SDK and scored 635 fps. Thanks to a graph compiled by the guys at WCCFTech, we can see that the RX 6000 cards are a bit better than the RTX 2080 Ti cards, yet they lag behind the RTX 3080 card by 33%.

Previous 3DMark benchmarks showed that the RX 6800 XT is around 20% slower than the RTX 3080, and we have to keep in mind that both the 3DMark tests and the DXR test are synthetic, so they do not really compare to how RT is actually implemente in games. Thus, AMD is probably not going to deliver Ampere-level RT performance, but at least the RX 6000 cards can be faster than the RTX 2080 Ti cads, which is no mere feat for a first try at ray tracing from AMD.

Estimated performance based on AMD's RDNA 2 info (Image Source: WCCFTech)
Estimated performance based on AMD's RDNA 2 info (Image Source: WCCFTech)
Read all 4 comments / answer
static version load dynamic
Loading Comments
Comment on this article
Please share our article, every link counts!
> Expert Reviews and News on Laptops, Smartphones and Tech Innovations > News > News Archive > Newsarchive 2020 10 > AMD sneaks additional details on the expected RX 6000-series ray tracing performance in the RDNA 2 overview
Bogdan Solca, 2020-10-29 (Update: 2020-10-29)