Numerical analysis on a solid state drive

About six months ago I reached a point in my research where my major models were taking over 24 hours to run. Convergence toward a working prototype was much too slow when iterating at this rate. I began digging under the hood and realized that my particular numerical code (a nonlinear finite element simulation) was writing massive amounts of data out to disk throughout the analysis.

Naturally I decided I should get a solid state hard drive (SSD).

This resulted in a ~4x speedup.

Will you get the same improvement? It depends, and there are many people who can answer that question better than I can. The basic line of thought goes like this: If your numerical algorithm is running primarily in RAM (or if you can increase the amount of RAM on your machine to encompass your simulation), then the SSD isn’t going to help you much. But if you see a lot of activity with your page file while running your model, you probably stand to gain a lot from moving it to an SSD.

For those using DIANA for finite element analysis like I am, the software developer provides a helpful discussion/example: Should I install a Solid State Drive (SSD) in my system to improve DIANA performance? — for the scenario they profile, the simulation sped up from 21313 to 7849 seconds (2.7x improvement).

An SSD is worth considering anyway for your primary OS because of how fast everything else will run. I had never used one before and was blown away. Firing up a huge FEA pre/post-processor on the old drive took about 15-20 seconds; on the SSD it’s live in 2-3 seconds. That makes for happy computing.

For those interested in my setup, I run Windows 7 (64-bit) on a 256GB SSD and have a second 2TB, 7200RPM storage/archive drive. I keep the SSD really lean—only hosting the OS and a working directory for active projects. Whenever I finish a project, I move it back onto the archive drive.

In summary: if you can afford an SSD for finite element analysis, you have very little to lose. It will definitely make your computer more enjoyable to use, and you may get some serious improvements in overall simulation time.

You can read more about me, follow me on Twitter, subscribe to this blog by RSS or email, and find many more posts in the archives.