Simulations of core-collapse supernovae in full GR on Exascale GPU supecomputers using the new GPU-accelerated dynamical-spacetime GRMHD code GRaM-X.

In this talk, I will present a new GPU-accelerated dynamical-spacetime General Relativistic Magnetohydrodynamics (GRMHD) code called GRaM-X which I have developed during my PhD. GRaM-X can perform production-level simulations of core-collapse supernovae in full GR on GPUs on world's fastest supercomputers. I will present the implementation, code tests as well as performance results (scaling) of the code. I will then discuss a jet-driven core-collapse supernova simulation in full GR that I have performed using GRaM-X on OLCF Summit and OLCF Frontier. I will present the results of the simulation as well as discuss the performance of the code in the simulation.