Now that we're uploading using a staging buffer,
we don't need to wait for non default buffers.
We should also respect READONLY for WRITEONLY buffers
(yes, it sounds dumb) because Nostale relies on that.
This parameter has three different meanings if it's non NULL:
- if Pool is D3DPOOL_SYSTEMMEM it's a host pointer for inital data
- if it points to a nullptr it's a output handle
- if it's a pointer to non NULL pointer it's an import handle
dxvk::high_resolution_clock is meant to be a wrapper around the C++ time
library, and getCounter() is (AFAIK) an internal thing so calling it
directly is odd. On winelib this redirects to the host's implementation,
which means this fails to compile. Since this is a Windows API and not
for internal usage just call the Windows API function directly.
This was only half-implemented (e.g. OpenXR was still calling
GetModuleHandle), broke compilation with Vulkan due to mismatched ABI,
and wouldn't have worked anyways with winelib builds because we still
need access to wine's implementation of the Vulkan win32 winsys
integration. Perhaps this is still useful for dxvk-native but if so it
should be re-added under a DXVK_NATIVE flag.
This isn't the meson way to do things. Doing this also prepares us for
using the builtin meson rc support instead of hand-rolling our own once
https://github.com/mesonbuild/meson/pull/8954 lands.
Aims to be mostly transparent to the application, although breakage
can still happen if shaders query the sample count and do not handle
a sample count of 1.
- Fix interleaved locks with evictManagedOnUnlock
We need to make sure there are no other subresources of a texture locked
before clearing the dirty box. Otherwise the data for those other subresources
won't get copied into VRAM.
- Clear dirty box regardless of texture pool
Otherwise we keep repacking and copying the whole texture
for every single lock. This causes performance problems
in Star Wars: The Old Republic.
If the buffer size is less than (vertexCount * stride), the last vertex
may be considered out-of-bounds, even if all attributes are included in
the vertex.
Fixes#2131.
We cannot invalidate mapped buffers in the backend because the frontend
may cache the mapped slice from its own invalidations, which would then
become invalid. Very unlikely to be an issue in practice.
The game constantly switches between SyncInterval 0 and 1, possibly in an attempt to implement adaptive Vsync, which causes performance issues because it forces us to recreate the Vulkan swap chain every couple of frames. Instead, enable the Vulkan equivalent of adaptive Vsync if possible.
Modifies GetSubresourceLayout slightly in case only a partial aspect
mask is being passed to the function. This way we can conveniently
compute the offset of a given pixel within the mapped buffer of an
image subresource.
