Home > News & Events > SDU Highlights Content
Shandong University's (SDU) School of Software achieved remarkable results at the 13th Parallel Application Challenge (PAC 2026), taking home one first prize in the Application Track and two third prizes in the Optimization Track. The awards were announced during the 22nd CCF National Annual Conference on High Performance Computing in Shenzhen.
In the Application Track, the university's winning entry introduced a unified high-performance optimization framework covering four types of force fields, ranging from classical empirical potentials to machine-learning potentials. By restructuring data flows, implementing multi-level parallelization, and enabling mixed-precision computing, the team overcame key memory and scaling bottlenecks in molecular dynamics simulations. They also developed a unified vectorization interface compatible with ARM, x86, and Shenwei architectures. This breakthrough enabled simulations involving tens of billions of atoms across millions of cores, providing solid technical support for large-scale materials research on domestic supercomputers.
Meanwhile, two graduate-student teams secured third prizes in the Optimization Track by tackling tensor contraction and Indexer operator challenges designed for scientific computing and large model inference. They developed computational restructuring and parallel execution strategies to reduce communication overhead and maximize multi-core utilization on domestic computing hardware.
PAC 2026 set a record by bringing together 862 contestants from 331 teams nationwide. The competition highlighted cutting-edge parallel application innovations and showcased SDU's growing expertise in leveraging domestic supercomputing infrastructure.
Application Track winners take a group photo. [Photo/Shandong University]
Optimization Track winners take a group photo. [Photo/Shandong University]