You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Salt particles can be treated either as atomic or molecular groups. For a speciation system with equal number of particles with and without nonbonded interactions:
group
ideal
nonbonded
atomic=true
0.4s
4.0s
atomic=false
12s
20s
Run using proposed code in #259 with this input: input.zip
Things to check:
Benchmark for bottlenecks
It is likely that the difference is in Space::findMolecules() for locating active/inactive particle. For atomic the call is more or less independent of N since the atomic group is ordered with active then inactive particles. For molecules, the operation scales with N. A Space particle tracker would solve this and was implemented in earlier Faunus versions.
Salt particles can be treated either as atomic or molecular groups. For a speciation system with equal number of particles with and without nonbonded interactions:
atomic=true
atomic=false
Run using proposed code in #259 with this input:
input.zip
Things to check:
Space::findMolecules()
for locating active/inactive particle. Foratomic
the call is more or less independent of N since the atomic group is ordered with active then inactive particles. For molecules, the operation scales with N. ASpace
particle tracker would solve this and was implemented in earlier Faunus versions.ping @rc83
Update: Removing the
sanity
analysis reduces the molecular time by a factor of six (ideal) or two (nonbonded). Atomic is still significantly faster.The text was updated successfully, but these errors were encountered: