IRIS Digital Asset - ATLAS GPF Machine Learning

The LHC experiments are collecting ever larger dataset, currently in excess of 1 EByte. To exploit the full physics potential offered by the large data sets, new software and algorithmic techniques are required to handle their production rate, volume, and complexity. The C++ code in use was developed during an era of entirely different hardware and software technology. The computing hardware has evolved towards parallel architectures, and scientific computing is moving towards the use of accelerator hardware, such as GPUs and FPGAs, where thousands of threads operate simultaneously to perform computations.