Yu-Chih Tseng @ on
simulation is slow for large diameters
I tried to run the simulator for fairly large InAs NW, with diameter ~5 to 20nm. Even with the number of conduction bands and k-points reduced, the simulation is still very slow, much slower than when D=2 to 4nm. Is there any reason for this?
2 Like 0 Dislike
Abhijeet Paul @ on
Hello The simulations can be slow depending on the venue for simulation chosen.
Presently we are working to remove that dependency. Most of the simulations presently run on a fixed number of CPUs, which makes it slower for larger cross-sections. A better version will have proper CPU increase as cross-section size increases. Also in case of job submission on clusters like steele, there is also sometimes long queue wait which adds to the total simulation time. At present we are working to achieve better CPU distribution and job-submission policy. Hopefully in the next version we will resolve this issue. Stay tuned.
Reply Report abuse
Please login to answer the question.