I can’t wait, because Id Software is about to drop Doom Eternal, the sequel to 2016’s surprisingly-faithful Doom reboot. Unfortunately, it’s not here yet, so we have to make do with the original for benchmarking. This game is unbelievably optimized for performance, and it shows in our results. We tested by running a set path through the beginning of the “Foundry” stage in Arcade Mode on Ultra-Nightmare difficulty using the following settings. As this test involves actual combat there is a bit of variability to the results, but it’s also completely representative of real-world game performance.
Doom is the only game in our testing set using the Vulkan API. As we’ve seen before, Nvidia’s Turing architecture absolutely loves the Vulkan API. The RTX 2080 Super turns in a handy win over the big Pascal chip on the GTX 1080 Ti, and the RTX 2070 Super is nipping at its heels despite being smaller and weaker on paper. If ever you needed evidence to prove to someone that Turing is a lot more than “Pascal-plus-RTX,” here it is.
For their part, the Radeon RX 5700 cards turn in a very solid performance. I suspect this game is tightly-tuned for the older GCN architecture, and while compatible, running in that mode takes little to no advantage of RDNA’s enhancements. As a result, we suspect Navi could do even better in this engine with a little tuning. Do take note of those 99th-percentile frame times. Likewise, check out those extremely skinny frametime plots for both Navi cards. Not only is their performance competitive, they’re extremely consistent, too.
All of the faster cards in this test put up astonishing performances. Here’s hoping that Doom Eternal is much more demanding than its predecessor, because Doom ’16 is quickly losing its utility as a benchmark. Even the humble GTX 1660 Ti put up an average framerate over 60 in our 4K test. The old Polaris-based RX 580 does particularly poorly here, but I suspect it’s being limited by its 4GB of on-board memory, as I mentioned before. Dropping down to 2560×1440 makes the game eminently playable even on that card.
These “time spent beyond X” graphs are meant to show “badness,” defined as those instances where animation becomes less fluid. The formulas behind these graphs simply total up the amount of time each graphics card spends beyond certain frame-time thresholds, each with an important implication for gaming smoothness. To fully appreciate this data, recall that our graphics card tests generally consist of one-minute test runs, and that 1000 ms is one second.
The 50-ms threshold is the most notable one, since it corresponds to a 20-FPS average. We figure that if you’re not rendering any faster than 20 FPS, even for a moment, then you’re likely to perceive a slowdown. A frame interval of 33 ms corresponds to 30 FPS, or a 30-Hz refresh rate. Go below that with vsync on and you’re into the bad voodoo of quantization slowdowns. 16.7 ms correlates to 60 FPS, the golden mark that we’d like to achieve or surpass for each and every frame.
In less-demanding titles, or when testing powerful hardware as we are today, it’s useful to look at our strictest graphs. 11.11 ms matches up to 90 FPS, a common refresh rate for ultra-wide gaming monitors. Approximately 8.3 ms is the frame interval at 120 FPS, the lower-end of what we’d consider a high-refresh-rate monitor, while most high-refresh rate displays now run at 144-Hz giving a frame interval of 6.94 ms.
On this title, we didn’t even bother including the 33- or 50-ms charts because there was nothing to show. In fact, only a few cards had any time beyond 16.7 ms (under 60 FPS) and really only one card stayed there for long: the aged RX 580 4GB. You have to drop down to 11.11 ms to start seeing any of the newer cards struggle. Incredibly, the RTX 2080 Super spends less than two seconds under 120 FPS in our test run. That’s sort of absurd.