I upgraded from a 3070 to 9070 XT about 4 days after the 9070 XT launched. Most games I am noticing what should be an obvious improvement in performance. With Dead by Daylight however, it's actually worse.
On my 3070, with Dead by Daylight set to ultra, Xess on quality and fps cap of 120 @ 1440p, it would be around 60 - 65% usage in a match. On my 9070 XT with same settings, it is about 85% usage. I've noticed that Xess even when on performance only reduces usage from native 1440p by about 3%, which is definitely weird. FSR has no effect on usage whatsoever.
I found a video of a 3070 Ti having better performance with this game: https://youtu.be/ndYL6EwBOcc As you can see it's around 60% usage on same settings I'm using except no scaling.
Here is a video of a 7900 XT at 4k with otherwise same settings and no scaling, doing 120fps lock with ~80% usage: https://youtu.be/3qA1KlYuepE This is a card that to my knowledge is ~ equivalent to a 9070 XT, and it is getting better performance.
The reason this matters is that eventually this game will get another graphics upgrade. It got its first about 5 years after launch, then last year it got moved from UE4 to UE5. When the next upgrade happens, if this performance issue isn't fixed, I will either have to lower settings to maintain 120fps, or drop to a lower framerate cap to maintain settings. When the usage, imo, should be around 50% or even lower with Xess on balanced, since Xess on quality on my 3070 with same settings otherwise was 60 - 65% usage for 120fps @ 1440p.
I have tried AMD Radeon Super Resolution - set the game to 1080p, then have RSR upscale to 1440p. This gets usage beneath 50% which I feel is where it should be if the game's builtin Xess scaling was interacting with the 9070 XT properly, however the UI is still 1080p so it doesn't look as good on a 1440p.
The solution I've found for now for low usage and headroom for maybe maintaining 120fps on very high or max settings when the next visual upgrade happens with Dead by Daylight, is using Fluid Motion Frames in the AMD overlay. I was worried this wouldn't work well with Dead by Daylight since any latency is very noticeable with the skillcheck gameplay, but it has been working fine - set the game to 60fps cap, turn on Fluid Motion Frames so it is 120fps (1 fake frame for every real frame), and I am still playing as well as I did without Fluid Motion Frames - which is to say no noticeable input lag. I also don't notice any image quality degradation, but maybe someone with a better eye for the differences would. Doing this lets me have max settings 1440p, no scaling (neither RSR through the overlay or FSR or Xess through the game) 120fps lock and less than 50% usage. If I turn on RSR I notice no difference in usage, Xess on even performance is still only a reduction of maybe 3% usage, so I am sticking with native.
Ideally this gets fixed properly so that there is even less GPU usage with or without AMD frame gen, so that there is even more headroom for future tech increases with this game. But for anyone who has been having a similar issue, it seems Fluid Motion Frames being set to on with the game capped at 60fps is the solution to high framerates and lower usage than older cards. If it matters, under the Fluid Motion Frames settings I have Search Mode on high and Perfomance Mode on quality.
I've sent in a ticket to AMD about this as well so they can hopefully fix it. I am optimistic that it is just a launch driver issue with a game they may not have tested. I have also tried the option March 20 driver but that didn't change anything.
tl;dr 3070 + Ti and 7900 XT have better performance with Dead by Daylight than the 9070 XT, unless you use AMD overlay scaling or frame gen, which shouldn't be necessary when the mentioned cards achieved better performance without these tricks.
