Intel oneAPI and Alibaba Cloud help Deepin Technology accelerate business applications

Deepin uses Alibaba Cloud's elastic high-performance computing and oneAPI to analyze and optimize the computing performance of the simulation platform .

DP Technology optimized the performance of simulation workloads by using Intel® hardware, oneAPI, and Alibaba Cloud Elastic High-Performance Computing (E-HPC) platform services.

Deepin is a global pioneer in artificial intelligence and molecular simulation algorithms. The company's Hermite® and Bohrium® platforms can significantly improve the productivity of various laboratories from the pharmaceutical and material industries to academia. For the LAMMPS (Large-Scale Atomic and Molecular Parallel Simulator) platform, the workload requires simulation of materials such as nanoclusters, metals, polymers, and dynamic processes such as growth, cutting, and fusion.

LAMMPS workloads are extremely challenging due to their inherent simulation complexity and ever-changing dynamics. To run LAMMPS workloads, Deepin Technology combined Alibaba Elastic High Performance Computing Platform services with Intel hardware. After optimizing with Intel® oneAPI DPC++/C++ Compiler and Intel® MPI Library, and profiling with Intel® VTune Profiler (available in Intel® oneAPI Base Toolkit and Intel® oneAPI HPC Toolkit), Deepin Technology improved the performance of LAMMPS workloads by 16.2%. After tuning the process and thread combination, the performance was improved by approximately 45.2% compared to the original Deepin Technology code using GCC and MPICH.

Tao Jinzhong, Product Manager of Alibaba Cloud High Performance Computing Solutions said: "Elastic High Performance Computing provides a fast, flexible and secure cloud computing platform for individual users, educational and research institutions, and public agencies. Combined with the Intel oneAPI toolkit, elastic high performance computing can help customers build an efficient analytical computing platform on Intel® Xeon® Scalable processors."