To help put this piece together, XFX sent me a factory overclocked example of the RX 6800 XT, specifically the Speedster Merc 319 edition of the card. Using its own tri-fan cooler design, this monstrous card delivers boost clocks rated at 2340MHz, around five per cent higher than the reference model, but I’ve seen it routinely reach clocks in the 2400MHz range and beyond. While the XFX card has new brand, the design philosophy has much in common with its previous ‘THICC’ range. The cooler is certainly similar, and does a good job in keeping temperatures at the low 70s Celsius range under load. As I mentioned though, it is somewhat beastly in term of form factor. Certainly in terms of its sheer bulk, you need to ensure you have adequate clearance in your case - it’s around 34cm or 13.5 inches in length! In terms of the actual ray tracing metrics, it’s best to refer to the video here for the full breakdown of how I tested individual RT effects and how well they are handled on each of our competing GPU architectures, but the key objective of this testing was to isolate individual stages of the RT pipeline to see how Nvidia and AMD perform, and to do so within the context of three key RT effects: shadows, reflections and global illumination. Typically, in any RT scenario, there are four steps. To begin with, the scene is prepared on the GPU, filled with all of the objects that can potentially affect ray tracing. In the second step, rays are shot out into that scene, traversing it and tested to see if they hit objects. Then there’s the next step, where the results from step two are shaded - like the colour of a reflection or whether a pixel is in or out of shadow. The final step is denoising. You see, the GPU can’t send out unlimited amounts of rays to be traced - only a finite amount can be traced, so the end result looks quite noisy. Denoising smooths out the image, and producing the final effect. So, there are numerous factors at play in dealing with RT performance. Of the four steps, only the second one is hardware accelerated - and the actual implementation between AMD and Nvidia is different, with GeForce cards having additional hardware grunt. RDNA 2 calculates ray traversal on the compute units, introducing competition for resources, while Nvidia does this in a specialised processor within the RT core. The first set-up stage may have significant CPU requirements, while the shading and denoising steps may have specific preferences for certain GPU architectures. For example, Quake 2 RTX and Watch Dogs Legion use a denoiser built by Nvidia and while it won’t have been designed to run poorly on AMD hardware (which Nvidia would not have had access to when they coded it), it’s certainly designed to run as well as possible on RTX cards. Regardless, in the video, I aim to be comprehensive in addressing the whole ray tracing pipeline on both architectures, covering off a range of effects. Ray traced shadows are tested in Call of Duty: Black Ops Cold War (an Nvidia-sponsored title) as well as Dirt 5 (supported by AMD). I take a look at ray traced reflections in Ghostrunner in Unreal Engine 4, where I can examine the effect with some degree of tweakability, and of course, Watch Dogs Legion’s reflections are also put under the microscope. I chose this one because AMD RT hardware is used in the consoles to deliver the effect, plus via modding, I can access both console and Nvidia denoisers. With ray traced global illumination, 4A Games’ incredible Metro Exodus is tested in depth, while I look at a more extreme example via the path-traced Quake 2 RTX - which now works on both AMD and Nvidia RT hardware, thanks to integration of finalised Vulkan RT extensions. So, what’s the takeaway? I think there are some intriguing results here. Ray traced shadows are generally inexpensive on resources on both the RX 6800 XT and RTX 3080 - with the RTX 3080 seeing minimal wins at lower settings, which then increase as the ray tracing quality is increased at higher settings, in a game like Call of Duty Black Ops. For ray traced reflections, the effect is much more demanding on GPU hardware, but the visual win is more pronounced in many scenarios. The higher the randomness of the reflected rays and the higher the amount of rays shot out, the greater the RTX 3080 fared in comparison to the RX 6800 XT, rendering in nearly half the time in certain configurations. The RTX 3080’s efficiency advantage lessened after a certain tipping point though, and I saw the same thing with global illumination: the RTX 3080 could render the effect in nearly half the time in Metro Exodus, or even a third of the time in Quake 2 RTX, yet increasing the amount of rays after this saw the RTX 3080 having less of an advantage. In general, from these tests it looks like the simpler the ray tracing is, the more similar the rendering times for the effect are between the competing architectures. The Nvidia card is undoubtedly more capable across the entire RT pipeline, and the RTX 3080 seems to have less dramatic performance losses as ray tracing complexity increases, but on the less complex end of the scale, AMD is competitive. Meanwhile, PlayStation 5’s Spider-Man: Miles Morales demonstrates that Radeon ray tracing can produce some impressive results on more challenging effects - and that’s using a GPU that’s significantly less powerful than the 6800 XT. And with that in mind, we do need to accept that ray tracing on the PC side is still in its early days, especially when running on AMD hardware. Right now, I can only deliver general conclusions from a representative, but still small sample. Thus far, we’ve only seen RT shadows in AMD-sponsored titles, and I’m eager to see how future titles developed in collaboration with Team Red fare on demanding RT effects. While ray tracing has been with us now in the PC space for over two years, the story is just beginning - and I can’t wait to see what comes next.
