Mirror’s Edge: Catalyst (DirectX 11)
Mirror’s Edge: Catalyst is not a great game, but it’s not bad, either. Because it wasn’t very popular, we figured it makes a fair choice for a highly-demanding game using EA’s Frostbite engine that hasn’t received the game-specific optimizations that a Battlefield title might. It helps that I own it, and I don’t own Battlefield.
We used the “Hyper” settings for testing. These came in a patch after the game’s release, and were specifically intended for the then-new Pascal graphics cards, which may explain their poor performance on AMD’s hardware. Playing this game on the RX 5700XT is a noticeably rougher experience than playing it on the RTX 2060 Super, despite the higher average framerate.
Our “Inside the Second” numbers bear that out, though. You only need check out the 99th-percentile frame times to understand just how bad things are for this game on the Radeons. With that said, the fact that this is notable is a good thing for AMD, because it means that’s usually not the case. Once upon a time we would have clucked our tongues and shaken our heads, but now we’re left scowling at the game itself.
Played on any of the GeForce cards—even the GTX 1660 Ti—this game is a delight and in this reviewer’s opinion, sadly-overlooked. I probably shouldn’t even have included the Radeon RX 580’s results, but this is yet another example of 4GB of memory being insufficient; at these settings, Mirror’s Edge: Catalyst wants around 5.5 GB of video RAM.
Our threshold charts further confirm what we already knew about this game: it runs like a dog on Radeons. Note that despite the higher average framerate, the RX 5700 XT spends nearly as long as the RTX 2060 Super on frames slower than 90 FPS. It also spends nearly the same amount of time on frames slower than 60 FPS as its own slower sibling, while the RTX 2060 basically spends none.
Keep in mind as you flip through these charts that the scale on all of them is a little wacky thanks to the Radeon RX 580. That card’s performance in this title is a significant outlier and probably should have been excluded.