Bio
Dr. Wenjun Ge is a computational scientist with expertise in building high-performance, scalable infrastructure for physics-based and data-driven applications. He specializes in developing GPU-accelerated, MPI-parallel software that runs efficiently on Leadership supercomputers. He has designed and optimized key components of exascale-ready software using CUDA, HIP and related math libraries, and has extensive experience in profiling, kernel-level tuning, and system-wide performance scaling.
He obtained his PhD from the University of California, Merced, in 2017. During his PhD, he developed the general mixed diffuse-specular boundary for the spherical harmonics method, and the first practical full-spectrum k-distribution look-up table for radiating gases in turbulent combustion simulations.
At ORNL, he has worked on multiple HPC projects developing GPU-accelerated software for scientific and engineering problems: including the spray and radiation modules for reacting flow, parallel strategy for DFT using MAGMA and NCCL, and BVH ray tracing. He also applied his knowledge and computational tools in reacting flow and radiative transfer to real-world applications, such as microwave-assisted CVI for high quality ceramic matrix composites, CVD for high-quality synthesis of 2D materials on wafers, and propane engines with fewer knocks and higher turbulence-dilution-tolerance. He views curiosity and the push to computing performance boundaries are the foundation to innovations in science and engineering.
Currently, he is collaborating with industrial partners and plasma scientists on diverse computational and data challenges that demand extensive computational resources.
Awards
2024 R&D 100 Finalist (Pele)