Copies tiles from buffer to tiled resource or vice versa.
void CopyTiles( ID3D12Resource *pTiledResource, const D3D12_TILED_RESOURCE_COORDINATE *pTileRegionStartCoordinate, const D3D12_TILE_REGION_SIZE *pTileRegionSize, ID3D12Resource *pBuffer, UINT64 BufferStartOffsetInBytes, D3D12_TILE_COPY_FLAGS Flags );
A pointer to a tiled resource.
Type: const D3D12_TILED_RESOURCE_COORDINATE*
A pointer to a D3D12_TILED_RESOURCE_COORDINATE structure that describes the starting coordinates of the tiled resource.
Type: const D3D12_TILE_REGION_SIZE*
A pointer to a D3D12_TILE_REGION_SIZE structure that describes the size of the tiled region.
A pointer to an ID3D12Resource that represents a default, dynamic, or staging buffer.
The offset in bytes into the buffer at pBuffer to start the operation.
A combination of D3D12_TILE_COPY_FLAGS-typed values that are combined by using a bitwise OR operation and that identifies how to copy tiles.
This method does not return a value.
CopyTiles drops write operations to unmapped areas and handles read operations from unmapped areas (except on Tier_1 tiled resources, where reading and writing unmapped areas is invalid - refer to D3D12_TILED_RESOURCES_TIER).
If a copy operation involves writing to the same memory location multiple times because multiple locations in the destination resource are mapped to the same tile memory, the resulting write operations to multi-mapped tiles are non-deterministic and non-repeatable; that is, accesses to the tile memory happen in whatever order the hardware happens to execute the copy operation.
The tiles involved in the copy operation can't include tiles that contain packed mipmaps or results of the copy operation are undefined. To transfer data to and from mipmaps that the hardware packs into one tile, you must use the standard (that is, non-tile specific) copy APIs like CopyTextureRegion.
CopyTiles does copy data in a slightly different pattern than the standard copy methods.
The memory layout of the tiles in the non-tiled buffer resource side of the copy operation is linear in memory within 64 KB tiles, which the hardware and driver swizzle and de-swizzle per tile as appropriate when they transfer to and from a tiled resource. For multisample antialiasing (MSAA) surfaces, the hardware and driver traverse each pixel's samples in sample-index order before they move to the next pixel. For tiles that are partially filled on the right side (for a surface that has a width not a multiple of tile width in pixels), the pitch and stride to move down a row is the full size in bytes of the number pixels that would fit across the tile if the tile was full. So, there can be a gap between each row of pixels in memory. Mipmaps that are smaller than a tile are not packed together in the linear layout, which might seem to be a waste of memory space, but as mentioned you can't use CopyTiles to copy to mipmaps that the hardware packs together. You can just use generic copy APIs, like CopyTextureRegion, to copy small mipmaps individually.