I don't like Denuvo either - breaking news, nobody does - but I'm allergic to breathless bullshit claims about anything and most of what's been said about Denuvo over the years has been exactly that. The only thing that was proven was prolonged loading screens, everything else is a pungent cocktail of sensationalism, hysteria and autism. And of course people trying to justify pirating games to feel better about it.
Pirates removed all traces of Denuvo from Asscreed Origins and there are considerable differences between performance there:
> Different time of Day
> GPU Utilisation is 100% on both sides
> Uncapped framerate
Something seems very fishy.
Denuvo is not known for running anything on the GPU, so I'd expect framerates to be consistent on both sides if the GPU is the limiting factor.
I've also written numerous posts around here about how games DO NOT SCALE LINEARLY, so even if you can say "my FPS is 25% higher", that doesn't mean denuvo's making the game 25% slower.
Games tend to be of the nature that requiring 5% extra CPU cycles can half your framerate or worse.
Something weird in that video is that the denuvo version has generally much lower cpu utilization and slightly lower ram usage. Vram usage is higher in that version tho.
Hard to say how he actually benchmarked.
But the RAM usage can probably be partly explained by windows just being a fat bloated pig, and him recording that footage after.
And VRAM somewhat by uplay's overlay being present.
CPU usage % is not a good way to know if your CPU is being well utilized at all.
Except CPU usage is the entire point, granted I much prefer the un-normalised number (i.e. I would personally see numbers ranging from 0% to 3200%).
As far as I am aware, DENUVO DOES NOT RUN ON THE GPU.
This means to benchmark it properly, you need to find a common baseline where BOTH sides push a smooth 60 FPS (or 120 or whatever), and then you compare the CPU Usage between denuvo and without.
As I have stressed repeatedly, GAMES DO NOT SCALE LINEARLY, and by turning it into a GPU benchmark, you invalidate the thing you're supposed to test.
And the results can then go in 1 of 3 ways:
1) No discernible difference
2) CPU Usage is lower - this is the expected result
3) GPU Usage is lower - completely unexpected result unless Denuvo does run on the GPU now.
and in the case of 2) or 3) we will finally have a figure for denuvo's actual impact.