Solver speed is directly associated with model accuracy and reliability. Finite element analysis comes down to solving very large matrices. JSOL has developed advanced techniques to solve these matrices faster and produce more consistent results.
JMAG users have some of the most advanced solver algorithms at their fingertips, and can be certain that JSOL will continue to adopt the latest calculation architecture within JMAG.
High speed processing and scalability are the two main goals of parallel processing. JMAG supports both shared and distributed memory processing and can incorporate the graphics processing unit (GPU) into calculations. Utilizing these architectures means that analyses with more than 10 million elements can be incorporated in a design cycle.
This case shows scalability of an MPP solver for stray loss analysis of a motor. With increase in parallels due to effective use of the core, we can see that speed is increasing in a linear fashion. In this example, the model is 3D and an eddy current analysis of approximately one million elements and 91 steps is performed. The calculation is completed in less than two hours using 128 parallels.