Determining the optimum resources for your simulation

When running simulations, the default setting is to use the maximum number of cores in the system. Depending on your simulation, this might run slower using all cores/processors. This page shows how to determine the optimum resource configuration to run your simulations. This will also come handy as a baseline when requesting for resources on the cluster to run your simulation job. 

Note: The most of the contents of this article apply to running FDTD in CPU mode. Information on running FDTD on the GPU can be found in the Getting started with running FDTD on GPU Knowledge Base article.

Check Simulation Memory Requirements

1. Open the simulation/project.
2. Check your simulation memory requirements. For simulations that do not have the memory check feature, skip to the next section, "Run with different resource settings". 
3. Make sure that you have enough RAM/Memory installed on your machine to run the simulation.

If your simulation requires more than the memory installed on your machine, expect the simulation to run slow since it will use your virtual memory (hard drive) to run your simulation. Otherwise, modify your simulation to lower the memory requirements or run the simulation on a different computer with sufficient memory.

Run Simulation With Different Resource Settings

1. Open Resources. 
2. Set Processes to about half of the number of cores on your computer, keeping Threads to 1. i.e. on a 24 core or more machine, 6 or 8 processes would be a good starting point. 

3. Save and run your simulation.
4. When the Status in the Job Manager indicates the simulation is done initializing and meshing and it is in the "Running" state, click Quit & save.

5. A log file will be created for your simulation, [filename]_p0.log in the same folder as your simulation file. Open the log file with a text editor. At the end of the file, it will indicate the FDTD solver speed in Mnodes/s.

6. Repeat the process and run your simulation while increasing the number of Processes, taking note of the solver speed for each run, until you are using the maximum available cores on your machine. e.g. 12, 18, 24, etc... (always even number).

Based on the results on the simulation log file, use the number of processes that provided the best performance (i.e. highest Mnodes/s).

Notes

• The number of processes will greatly depend on your simulation/model and not just on the hardware and could vary for each simulation and computer used.
• This process can be used to compare the FDTD solver speed between CPU and GPU.
  • Run the simulation/model on CPU and obtain the best resource configuration for the fastest FDTD solver speed (Mnode/s).
  • Run the model/simulation on GPU and compare the FDTD solver speed (Mnodes/s). 

See also

FDTD Performance Benchmarks 

Information on Hardware Specifications