Introduction to AAD-Compiler from Matlogica at Intel HPC Workshop

Dmitri Goloubentsev from Matlogica presents a new much faster way to perform AAD (Automatic Adjoint Differentiation). This technique also allows to enable safe multi-threading and vectorization for generic programs where multiple data samples have to be processed in parallel.
As a result, AADC from Matlogica can turn your single-thread object-oriented analytics to parallel HPC world and you can get all sensitivities at the same time.