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
Initial tests of the CLI in #972 on a 50K atom system are showing very poor performance with replica exchange.
Even when correcting for the number of replicas via the fix for #999, I'm seeing only ~120 ns/day on an RTX 2080.
We should compare with the unmodified system and see what might be causing performance degradation here.
The text was updated successfully, but these errors were encountered:
It looks like the logging from the replica exchange simulation is suppressed by default?
I'm particularly interested in looking at the timings per part of the iteration cycle.
Initial tests of the CLI in #972 on a 50K atom system are showing very poor performance with replica exchange.
Even when correcting for the number of replicas via the fix for #999, I'm seeing only ~120 ns/day on an RTX 2080.
We should compare with the unmodified system and see what might be causing performance degradation here.
The text was updated successfully, but these errors were encountered: