GRMHD Simulation Framework
A high-performance parallel code for GRMHD simulations on GPU clusters.
We build a scalable GRMHD pipeline including discretization, multi-GPU parallelization, and validation on canonical test problems.
- Implemented high-order finite-volume solver with robust shock capturing.
- Achieved near-linear strong scaling up to N GPUs on cluster X.
- Validated against standard benchmarks (e.g., shock tube, MRI, etc.).