adcc.get_n_threads() doesn't seem to work
EvGeGr opened this issue · 2 comments
Hello,
I noticed that adcc.set_n_threads() doesn't seem to affect the number of threads, though adcc.get_n_threads() gives the number of threads setup.
Just a curious question, how efficacy (speed) of adcc is as compared to that in QChem?
Kind regards,
Evgeniy
Hi,
what lead you to the conclusion that set_n_threads(num_threads) does not affect the number of threads? adcc should definitely run in parallel setting that flag, at least for the officially deployed versions 😄
We did some speed comparisons when we wrote the paper (I think it's in the SI), but have since improved performance of some routines by a bit, but I don't have any benchmarks at hand.
Hi,
many thanks for your response! Concerning the number of threads, I meant that I noticed (when monitoring an adcc job with top) that adcc always runs (at least in my case) with the maximum available number of threads (CPUs) despite setting set_n_threads() to a particular (less than maximum) number of threads.
I also noticed (again using top) that my addc job (the python process) is always in "S" (interruptible sleep) state. Is that normal?
Kind regards,
Evgeniy