Employing HPC for Heterogeneous HEP Data Processing

The DEEP-EST Project aims to build a Modular Supercomputer Architecture (MSA) with the main focus on the usage of heterogenous computing resources. The majority of current HEP workflows are capable of utilizing exclusively Central Processing Units (CPU), which create inefficiency when trying to run on heterogenous systems equipped with additional hardware accelerators (GPUs, FPGAs).
Within the context of this project, we want to extend current CPU specific implementations of various algorithms (calorimeter reconstruction, track fitting, etc. ) by utilizing OpenCL and CUDA language extensions in order to harness the computational power provided by various hardware accelerators.