In the Summer of 2022, I worked as a Graduate Research Aide (another name is called "internship") at Argonne National Lab under Dr. Chris Knight and Dr. Alvaro Vazquez Mayagoitia to develop molecular simulation software compatible with graphic cards. 

Building upon our in-house serial code RASPA, I wrote a CUDA/SYCL/C++ software that parallelized configurational-bias Monte Carlo and Ewald summation in Monte Carlo simulations aiming for small systems with the size of smaller than 10,000 atoms. We also compared and tested the speed of different ways of parallelism for small systems in molecular simulations. 

** The software is called gRASPA (pronounced as gee-RASPA). 

Here are some speedup comparisons for Ewald summation for creating a CO2 molecule in 8 MFI unit cells (2,304 atoms).