July 2, 2021 10:17 PM, "Henri Verbeet" hverbeet@gmail.com wrote:
Or is this succeeding because we fall back to a separate allocation in vkd3d_bind_heap_memory()? In that case though:
- Why is that a problem for the actual application?
It's a problem not because of any failure but because heap memory is wasted. That's the only reason for the patch.
- We may be able to detect that by aliasing multiple resources to
the same heap offset.
Would this work by placing a smaller resource at a larger alignment within the same heap area, and check if its data has already been set to the uploaded value?