SGI Technology Guide for ANSYS Mechanical Analysts
ANSYS Mechanical tackles all important Normal Mode Analysis utilizing either Shared Memory Parallelism (SMP) or Distributed Memory Parallelism (DMP). The technical approaches used here include block lanczos and supernode eigensolver. Efficient execution of the above two paradigms and solutions requires extreme care in allocating hardware resources. The topic is even more important given hardware variety today, ranging from single node multi-core workstations through multiple node clusters to single image many-core systems addressing very large memory space. The paper will explore trade-offs in computer resource requirements of such solutions to establish guidelines for advanced SGI computer hardware systems.