Tests and Benchmarks for TMVA/SOFIE

TMVA/SOFIE is a new experimental module part of the ROOT project that is intended to convert pre-trained deep neural networks into a C++ header file that carries minimal dependencies on external libraries. This report is intended to summarize the work that I did as a participant to the CERN Online Summer Student Programme 2021, held remotely. The outcome of this work is code, as this is mainly a software project. I organized a CMake solution to allow quick and easy integration of GoogleTests and GoogleBenchmarks into the SOFIE project. I also developed a class that allows the user to profile an inference task in SOFIE via code instrumentation. This is helpful in identifying the computational bottlenecks of the generated models