Robin Kertels
Robin Kertels
Looks like an issue with a depth prepass. That buffer error is definitely interesting too. I'll take a look tomorrow.
I'm not seeing any issues when replaying the apitraces on Nvidia. EDIT: Enabling strict float emulation makes it show up on Nvidia too.
Nothing is gonna happen here until you make an apitrace.
Doesn't always using MTLHeap for VkDeviceMemory also require a rework to do explicit barriers? Metal tracks the underlying heap of each resource, so it would lead to a lot of...
Maybe we should start considering just always using cached memory for `D3DUSAGE_DYNAMIC`. @Joshua-Ashton This issue seems to be pretty common for D3D9 games.
Needs to be re-implemented first.
Maybe start by naming a particular game. Then tell us whether the problem is CPU performance of GPU performance (is GPU usage 100%?) In my experience DXVK runs Source games...
@JohnnYDeveloperAHK That misses the point. The bug report points out that GPU performance is inferior to regular D3D9 on Windows or ToGL. > because if it STILL does reach constant...
I'm getting a bit confused here so I'll try to sum it up so far: The initial set of screenshots (https://github.com/doitsujin/dxvk/issues/4895#issuecomment-2849591749) were done on a GTX 1650. DXVK doesn't manage...
> i think dxvk 1.10.3 has about the same performance as the older version. DXVK 1.10.3 is ancient too. > no, this is indeed without a doubt being ran on...