An OpenMP-Based Parallel Execution of Neural Networks Specified in NNEF

Recently, we have many research works on the neural networks and their related issues. For exchangeability of neural network frameworks, the Neural Network Exchange Format (NNEF) specification is now widely used. Due to very large size of these neural networks, their accelerations are actively explored, and can be achieved through parallel processing techniques. In this work, we present a prototype implementation of C++ code generator with parallel-processing accelerations based on OpenMP, for the NNEF specification files. Our implementation shows remarkable accelerations, in comparison to the original C++ template-based execution. We will tune the prototype acceleration to achieve more remarkable speed ups.