SIMD acceleraton for RB dihedrals
The RB dihedrals are not SIMD accelerated and can take a substantial amount of runtime. Adding SIMD codepath for these kernels would improve performance not only in exotic cases like the polymer system discussed on gmx-users which has only RB bonded interactions, but also e.g. in membrane systems with Berger lipids.
SIMD acceleration for RB dihedrals
RB dihedrals now use SIMD acceleration analogous to proper dihedrals
when no energy and virial is required. This also significantly
improves load balancing (issues) for systems with proper+RB dihedrals.
#2 Updated by Mark Abraham over 2 years ago
I hope to be able to do this as part of my bonded-stuff upgrade, but people reviewing C++ conversion and https://gerrit.gromacs.org/#/q/status:open+project:gromacs+branch:master+topic:bondeds will speed this up.