https://wccftech.com/devil-may-cry-5-pc-performance-explored/

Capcom is at it again, this time with Devil May Cry 5.  Unlike Resident Evil 2 this one isn’t a reboot but a full fledged follow up to a very popular series.  In this one we’ll be looking Devil May Cry 5 PC performance and thanks to the fact that it’s running the excellent Reach for the Moon Engine (RE Engine), that Capcom has been capitalizing well off of since the launch of Resident Evil 7, we have a pretty good indicator of what to expect with the results in this one.

Testing this game was more than a little bit challenging.  Not because it lacks a benchmark utility, that’s something that is easy to work around with a little bit of planning, but rather the swing in performance through scenes.  Reminiscent of Metro Exodus where framerates can go from over 100 to below 60 by simply looking the wrong way.  The game also performs very differently based on which area of the game you’re in.  But the biggest monkey wrench in the whole thing comes in two forms; multiple variable settings and whether the game decides it wants to launch in DX11 or DX12.  For our testing we took our results from the main intro cutscene which uses the game engine and we allow it to run the course from the moment the Capcom logo disappears to the time that the van crashes or roughly 222 seconds.  Settings wise we used the highest settings available UNLESS they were variable, the settings that are set to OFF were those which only allowed for variable settings and due to the very nature of variability, we left them disabled to maintain consistency.  The API debate is an interesting one here as the game game supports DX12 but features a fallback to DX11, more often than not it launched in DX11 for us so we set that to the target API.  There is not an option in the game to do this so you’ll have to rely on the .ini file.  But another thing to take note is that even in the scene we used the frame rate varied wildly and resulted in a pretty wide disparity between averages and 1% minimums so evidences of stutter will need to be seen in the spread from 1% and .1% minimums. And much like the results Metro Exodus the performance metrics here can be used to have an expectation of performance with it often higher and lower but is a good indicator of performance variation between CPU cores, API, and graphics cards.