Support multithreading for charge drift simulation

[gabrigarjim]

When simulating the drift of a charge cloud with

drift_charges!(evt, sim, Δt = time_step, max_nsteps = 50,diffusion = true, self_repulsion = true, verbose=true)

simulate!(evt, sim)

I don't see any improvement on the performance by previously setting

export JULIA_NUM_THREADS=10

to 10, 1 or any other number of threads. I do get a faster calculation in the case of the potential and weighting potential calculation, but not here. Is there any extra argument or step that I am missing?

Right now I am getting rates of 1 event/second for a highly segmented detector. As you can imagine, simulating 1E6 events becomes painfully slow...

[oschulz]

At the moment, the charge-drift simulation is not multithreaded. We are currently doing a major rewrite of it, trying to make it fully parallel and ideally GPU-compatible, that will still take some time though. In the mean time you can just simulate_waveforms(...) mulitiple times in parallel yourself (it should be thread-safe) on multiple Julia threads.

[gabrigarjim]

Oh I see! I will do as you say then. Happy coding!

[oschulz]

Thanks! We'll keep this issue open until things are running in parallel.