Correctly clear fragment shading rate attachments with LOAD_OP_CLEAR
Fix potential device lost with vkCmdBlitImage when fragment shading rate is used
Correctly restore pipeline static state if previously overwritten by dynamic state set calls and pipeline is rebound
Fix dual-source blending when output is written to an array variable
Fix VK_KHR_cooperative_matrix bug with variable pointer shared memory CTS tests
EDIT: if I had forspoken I would’ve just tested it, but I don’t have that game. If you can install that specific driver use VKD3D_CONFIG=skip_driver_workarounds to allow execution of the specific extension on VKD3D then it should work as intended. But note 535 will not allow you to play starfield.
Yes, you’re right - I only tested Starfield with VKD3D_CONFIG=skip_driver_workarounds and 535.43.09. I’ll try Forespoken later for here is 3 AM and I’m falling asleep.
The game doesn’t run with 535.43.09, VKD3D_CONFIG=skip_driver_workarounds and Linux Kernel 6.4.12. The game does run with 535.43.09, VKD3D_CONFIG=skip_driver_workarounds and and Linux Kernel 6.5.1-273-tkg-cfs but there’s no difference at all - the perf is terrible.
Thanks for the update! Perhaps it’s best to be checked yet again - as the suspicion from VKD3D devs was that extension was causing the issue. Now that it’s resolved then the issue most likely lies somewhere else(for both Halo Infinite/Forspoken, Starfield is mixed results so far to confirm).
Hi All,
Issue is not root caused yet, but I can see update on the bug on regular basis, team is actively working on it to fix it.
Thanks for being patient.
Does passing SteamDeck=0 %command% makes any changes to perf situation? Yes, it can add a flickering but mainly i want to know if it makes any perf diff.
As it seems to be forced as =1 to everyone currently, to mitigate some issues but maybe has some side effects on NV setups.
