These resource types may cause inefficient partial writes over the Xᵉ-LP 64-byte cache lines. Avoid these partial writes to get maximum bandwidth through the cache hierarchy. This can be done by ensuring that a single thread executing a given shader on a 4x2 group of pixels writes a contiguous 64 bytes on its own for output.
Access to read-only data is much more efficient than read/write data. Use these kinds of resources with caution and when there are no better options.
Do not set a resource to use a UAV bind flag if the resource will never be bound as a UAV. This programming behavior may disable resource compression.