The First Battlefield V DXR Ray Tracing Benchmarks Have Dropped

Ever since Nvidia announced the RTX GPU family, we've been waiting to find out what performance would look like in shipping titles. The demos on-hand at the launch event were widely reported to show even the massive RTX 2080 Ti struggling to maintain 60fps in Battlefield V, which didn't bode well for the performance of lower-end GPUs.

Because Nvidia raised prices significantly with the RTX family of GPUs, there have been ongoing questions about what kind of performance these cards would achieve in RTX games, how fast they'd be, and whether cards like the RTX 2070 would be capable of maintaining playable frame rates in DXR-enabled games at all. (DXR is the name of the official DirectX ray tracing standard from Microsoft, RTX is Nvidia's name for its specific implementation). The DXR implementation of Battlefield V has finally dropped, and the earliest benchmarks and visual comparisons are popping up. TechPowerUp, who covered the game's non-DXR performance last week, has repeated their tests(Opens in a new window) on this new version as well. Battlefield V(Opens in a new window) doesn't just offer a single On/Off switch for DXR -- there are specific detail levels, including Low, Medium, High, and Ultra.

Here's an example of how the game looks at low DXR detail versus Ultra. Unfortunately, TechPowerUp didn't take screenshots with DXR disabled, so we can't offer those for comparison.I've picked these two scenes to showcase in a comparison because, while they don't show the DXR ray tracing to its shiniest effect, they do show some very interesting things about the way ray tracing is being computed in-game. See how the metal bars in the train car are unlit in "Low," but illuminated in Ultra? This suggests "Low" is missing bounces. The transparent glass isn't actually showing light the way it ought to. You can also see distinct differences in how the light is playing off the building behind and to the right of the trolley car. The Low version looks almost more like a DX11 god ray effect, while the Ultra ray tracing shows a much more realistic lighting model.

The decreased number of bounces is probably responsible for one of the performance differences we see in the actual test results themselves. I don't want to show too much of TechPowerUp's work, but in the interests of giving a different example of how good ray tracing can look, here's "Ultra" quality in a different scene. For more details and comparison shots, check the original story:

The RTX Performance Question:

TechPowerUp's benchmark results are, if I'm being honest, kind of weird. Medium, High, and Ultra results are clustered tightly for the cards in question. Only "Low" performance enables RTX at a significant performance improvement relative to these other settings. Because 1080p is the resolution that everyone has talked about, we'll focus on it -- 1440p and 4K results are available at TechPowerUp for comparison.

The RTX 2080 Ti is capable of maintaining 65fps at 1080p, compared with 54fps for the 2080 and 47 fps for the RTX 2070. That's ironically exactly where we thought the RTX 2070 would land, though in our original example we assumed it would be the RTX 2080 pushing up to 60fps at 1080p with the 2080 Ti landing above that point. Performance is a bit compressed relative to what you might expect on RTX-Ops alone. The RTX 2080 Ti has 1.78x more rated RTX-Ops than the RTX 2070, but it only performs 1.39x better. Whether this reflects low efficiency or a bottleneck elsewhere in the GPU is not clear.

The Enthusiast Paradox of RTX

For years, AMD and Nvidia have pushed gamers towards a specific set of upgrades. Gaming was to be enhanced via higher resolutions, better color reproduction and HDR, the adoption of G-Sync / FreeSync, and higher frame rates. The PC gaming enthusiast community has generally jumped aboard all of these trends, with some niche interest in multi-monitor gaming or 3D gaming a few years back.

Nvidia's RTX implementation turns many of those assumptions on their heads. Now, better visuals are being delivered through a new rendering method at the expense of many of the features Nvidia previously stated were fundamentally core to being a high-end PC gamer. If RTX cards(Opens in a new window) had debuted at the same price points as their GTX predecessors, I don't think this would've been such a controversial position to take, but gamers are essentially being asked to fork over even more money to play games at lower performance levels. The more you invested in the 4K / high refresh rate ecosystem, the more you spent on features that won't be used in RTX games.

It's entirely possible that enthusiasts will simply shrug and say they've invested heavily to have access to the best of both worlds in every scenario. I don't personally play in the rarefied air of the ultra-high-end gaming market. But I think part of the backlash against RTX has been precisely that people who invested in pushing the edge of gaming in very different ways are now forced to idle those capabilities if they want to use RTX in the first place. The people RTX seems best-suited for are those with lower resolution displays and slower refresh rates in the first place -- and while that definitely represents the bulk of the market, most gamers don't invest in $500+ GPUs either. But regardless of how you slice it, gaming in RTX imposes a substantial penalty, and only the most well-heeled gamers are going to be playing in resolutions above 1080p. The performance impact of RTX looks to be heavy indeed.