How Can HPC Speed Up my Simulations?
Ansys Lumerical products are not only capable of utilizing all of your local computer's or workstation's resources to the fullest, but they can also utilize the resources of multiple machines (a cluster) to drastically reduce your "time-to-results" with high-performance computing (HPC).
There are two main ways that a cluster can be utilized to speed up your simulations (as well as a 3rd hybrid configuration):
- Concurrent Parametric Computing: Running multiple simulations in parallel (ie. sweeps and optimizations).
- Distributed Computing: Running a single simulation, spread across multiple nodes (for FDTD and varFDTD simulations).
See FDTD Performance Benchmarks for information on the performance of FDTD on some of the AWS instances we have tested on. See Information on Hardware Specifications for details on how different hardware components affect Lumerical FDTD simulations.
How Can I Get Started with HPC?
Using cloud and cluster computing services is a great way to increase the speed and throughput of your workflows.
Amazon Web Services (AWS)
Use our AWS EC2 services guides to easily start running your simulations on AWS. These guides will help you identify the type and amount of computing resources necessary for your workload. You can check out the AWS EC2 documentation for more information.
Local Clusters (on-premise)
This page in our KB shows the process of configuring a local on-premise cluster and running Ansys Lumerical simulations on the cluster.