|
简单来说,就是 Compute Shaders on D3D10 Hardware,是 DX11 给 DX10 做的补遗。
和足本的 CS 5.0 相比,CS 4.X 缺少 Atomic operations、Append/consume、Typed UAV access (unordered access view)、Double precision、DispatchIndirect() 的支持;
以及 Only a single output UAV allowed、Not a huge restriction in practice、Thread group grid dimensions limited to 65535、Z dimension must be 1 (no 3D grids)、Thread group size is restricted to maximum of 768 threads total (1024 on D3D11 hardware)、Thread group shared memory restricted to 16KB total(32Kb on D3D11 hardware)。 |
|