Expand description
Reference CPU implementation of a compute layer.
This implementation is not optimized to use multi-threading or SIMD arithmetic. It is optimized for readability, used to validate the abstract interfaces and provide algorithmic references for optimized implementations.