Reshade is just more of a hassle to me. It’s so quick and easy with the freestyle filters. I also haven’t noticed any significant impact on performance with them ever.
Yeah, it occurs on older drivers as well, a friend just let me know. That's too bad. I got two games I wanted to play, Hogwarts Legacy and this one. The first has its lighting system completely messed up after the latest patch, the latter crashes randomly. 2023 looks prosmising hehehe. I am having more fun playing Quake 2 rtx right now than anything else. It doesn't crash and it's a lot of fun.
Personally I had a 10-15% lower framerate on the 3080 after activating one of the filters. On Reshade it's 2-3 fps. Come on it's not a hassle. Actually it's pretty straightforward and works on anything almost.
It may not be a huge hassle but it’s still like 1000% more work than the filters. The less I have to do the better. 10-15% is wild. Even 2-3fps is more than my use of the sharpen filter on my 3080 causes.
Tested Novigrad (my sweet spot) and it's getting +7 to 11FPS more. Glad they did something. Now running in Novigrad is a looot less stuttery.
Still running great not a single stutter in 2 sessions. CPU (13600k) is also getting some nice usage. 4k Ultra no RT 7900xtx
I see CDPR fixed the CPU issues in the last patch. Right... That solved everything! Congrats... Pats on their backs and crack open the complimentary champaign... DX12 : 41 fps DX11: 79 fps DX11-> Vulkan (DXVK 2.1) : 94fps Same settings (RT always OFF!), pretty similar graphics quality (there might be very slight differences but it's hard to tell with the dynamic weather and it's definitely not something that warrants a ~50% performance hit, not even 5%...). I would expect a sane DX12 port to perform at least as good as DX11, and I would expect a decent DX12 port to perform as well (or even better) than the DX11->DXVK wrapping (which is effectively the laziest and most ineffective way of converting DX11 into Vulkan). P.S.: This is the first time I tried DXVK with TW3 and I am honestly surprised I found the first game which runs significantly faster with it on my system. I have never seen anything like this before. DXVK usually gave me the exact same fps as native DX11 in all the other games I tried (but usually with more stuttering when the shader cache was cold). I wonder if this is a TW3 thing or DXVK came a long way with it's updates and I shall re-test some other DX11 games as well. Too bad I couldn't find compiled DLL's for the DX12->Vulkan wrapper (though I am skeptical if that could improve anything with the DX12 port).
Imho Nvidia have mostly stopped optimizing their D3D11 driver, perhaps years ago. In Sons of the Forest, AMD D3D11 driver can handle overhead way better. I was getting absurd >80% CPU usage on 11400F when walking into some new areas. With 6700 XT, it sits at ~45% despite much higher fps.
Yeah, I am not sure what's going on (at large) because I also saw this video where somebody with an nV RTX3050 VGA and an AMD 5600X CPU (so same architectures but lower performance tiers there compared to my machine) had a minimal performance difference between DX11 and DX12. . And that even was before the patches which improved the CPU side of the DX12 things. Yet I can also find other people complaining about a ~50% difference using google, look here for just a few examples: https://steamcommunity.com/app/292030/discussions/0/3731826212313632146/#c3731826212314710353 (there are others). It's strange how this phenomenon wasn't inspected by the likes of the big Youtube channel operators. I opened a new thread for my findings: https://forums.guru3d.com/threads/strange-dx11-performance-issue-win11-22h2-ampere.447161/
@janos666 Maybe it helps, I remember I had this issue before the last patch, hence I was playing the game in DX11. There was considerably more fps in DX11 and less stuttering. But not anymore. Could you try to disable HT on CPU and see if it makes a difference?
There is still a massive difference for me between dx11 and dx12 with identical settings, with newest driver and newest game version. DX11 4k ultra (no dlss) DX12 4k ultra (no dlss) DX12 4k ultra RT (Quality dlss) RT included as a bonus. DX11 95 fps vs dx12 71 fps... that's a HUGE difference. Notice how much more vram dx12 also uses for no apparent reason.
I tried disabling SMT (first through Task Manager's affinity settings, then at firmware Setup level). It didn't change anything at all. (That was expected because most of the workload was already scheduled on ~2 CPU cores. Nothing was "pinned in a bad way".) That isn't necceserely a bad thing. That RAM is there to be used (it's probably caches more assets in VRAM).
Im all for caching, but it obviously isn't making dx12 running better than dx11. But even if they do somehow fix the cpu usage with dx12, the dx12 version still uses a whole lot more gpu aswell.
Found it! I can replicate the fps loss on DX12 using the same settings. The problem is in my case: Screen Space Reflections If set to High on both DX11/12 I get like almost half the fps on DX12. If set to Low on both I have almost the same fps. And by the look of things, setting it to High in DX11 doesn't change anything? Maybe that's why there is no fps loss vs Low.
