First milestone of the #openEMS high-performance engine rewrite project: empty box simulation is busy multiplying zeros by zeros running on 56 threads at 1 gigacells per second on a dual-socket Xeon E5 v4. The code is completely useless at this point, it's not even really faster than the upstream engine. which is actually pretty fast when there's nothing inside the simulation box. But I'd still call it a win since it didn't crash even once during initial system bring-up when I tried integrating my test kernel into the actual source tree. #electronics
Thanks to the recent work by Tobias Ammann, I've confirmed today that realistic dielectric material modeling of circuit board substrate is now almost a solved problem in #openEMS. The next problem for the community to tackle is realistic conductor modeling, with corrections for surface roughness and surface finish. #electronics
Just converted Mur's absorbing boundary condition in openEMS to use loop tiling as well. Speed of my simulation jumped from 172 million cells per second to 1000 million. 📈
@niconiconi what are you using #openEMS for, modeling radio interference between the components on a PCB, RF/analog design of PHYs (or macrocells for ICs), something else?
My preliminary result of implementing diamond tiling optimization for #openEMS - 2x to 6x speedup. The GCPW example saw a massive 600% speedup, time-to-solution reduced from 1 hour to 10 minutes. It's a very small PCB surrounded by the Perfect Matched Layer extension of comparable size, in this case the overhead of poor data locality and thread synchronization seemed to be especially high. #electronics#hpc