* In heavy D3D11 workloads the refcounting overhead especially during fast
binding changes was significant. Refactoring the refcounting to work on a
different model and deferring destruction of objects removes most of the
overhead.
* This helps catches cases where a discarded image is accidentally used and in
many cases may still have valid data. Particularly on Vulkan this is relevant
for DONT_CARE renderpass load and store ops.
The rules for merging semantics into an array were not strict enough.
If either the type, interpolation mode, or size is different, or if it
is not using a register's x component, then we can't combine it. Also,
the rules for marking a semantic as array length 1 were too strict,
resulting in some semantics packing into other registers incorrectly.
When a resource is accessed, it is now tracked by the debug step. The
shader viewer has a new panel to display resources accessed up to the
current step, with context menus to go to previous/next access of a
specific resource.
When getting read only resources from the pipe state, stitch back up
according to the bindpoint mappings. When displaying resources in UI,
don't traverse unbounded arrays. Fix resource swizzle on load/sample/
gather instructions, which happens on fetch result, not on the source
operand. Added more tests for unbounded arrays and different ways to
index into arrays.
The assignment of logical identifiers is not guaranteed to match the
order of appearance in the shader bytecode. Adjusted D3D12_CBuffer_Zoo
to exercise this for testing.
Added fetching of texture data for UAVs, to allow stores to occur.
Fixed issues with sample/gather when debugging a compute shader.
Added a demo for dispatch in D3D12.
SV_PrimitiveID is only valid as a pixel shader input if there's either
no geometry shader, or the geometry shader emits SV_PrimitiveID. With
D3D11, the debug layer will complain but rendering will proceed. With
D3D12, the PSO fails to compile. For both APIs, pixel shader debugging
now checks whether this is true, and disables its usage if it isn't.
Added tests for various valid scenarios with SV_PrimitiveID.
The logical identifiers, rather than shader registers, were used to
populate reflection info, causing the wrong slots in large descriptor
tables to be referenced for the pipeline state.
* It's not particularly scalable and can be brittle to driver changes, and we
can use targeted specific pixel tests to check what we really want - to see if
the output has rendered correctly.
* Overlay tests still check files directly - this is a future refactor to
remove.
This mimics the D3D11 shader debugging test, but performs the draws
twice - once with SM 5.0 and once with SM 5.1. The python test performs
the same checks as the D3D11 version, but is explicitly disabled since
D3D12 shader debugging is not yet enabled by default.
baldurk