We will use this extension in order to implement vertex
binding divisors other than 1 for per-instance attributes.
Will be *required* as soon as support by wine and Vulkan
drivers is widely available.
We no longer use the 'oldSwapchain' member in the swap chain description
to replace an existing swap chain, but rather destroy it altogether and
create a new swap chain. While this is less than optimal, it might help
solve some swap chain-related issues such as #277.
An alternative to manually creating a framebuffer object and binding
it via bindFramebuffer. Future optmizations can use this to bring
down the number of redundant render pass changes.
If this method is used to clear a view with a floating point format,
we need to create a compatible view with an integer format in order
to clear the resource with the correct value. Fixes some calls to
this function in Rise of the Tomb Raider and other games.
The D3D11 ClearUnorderedAccessView* and ClearView functions
will have to be emulated using compute shaders rather than
clear operations, since Vulkan clear operations do not take
image views and their format into account.
We don't need to iterate over the full shader code when creating
a new shader module. This optimization may slightly reduce the
initial pipeline creation time.
Drivers from both major vendors implement their own shader cache
already, and storing a cache per game causes more issues than it
solves. Should fix#261.
Since we are synchronizing once per frame anyway, there is no need to
artificially limit the number of chunks in flight. Applications which
use deferred contexts and submit a large number of CS chunks through
command lists may benefit from this optimization.
HUD elements can be enabled individually using a comma-separated
list. Supported options include:
- fps: Displays the framerate
- devinfo: Displays device info
Passing "1" has the same effect as "fps,devinfo".
Since we create only one DxvkContext per D3D11Device, rather than
per D3D11DeviceContext as originally planned, there is no need to
keep the pipeline manager as a global thread-safe object. This may
slightly reduce CPU overhead.
