HanatoK

> Taking a step back, since we are purely talking about optimization here, I'd like to start from use cases. What expensive CVs would users really like to use? I...

> In the cases you mention, it seems there are fewer CVCs in total than cores on a typical CPU, so with our current scheme, the time to complete a...

I want to emphasize that even in case of calculating two slow CVs (RMSDs for example) with two CPU threads, it would be better to parallelize the inner loops instead...

We have discussed this issue in https://github.com/Colvars/colvars/pull/780. The problem is that once we document it, we cannot change the name `inner_loop` anymore. However, according to @giacomofiorin's comment, the name `inner_loop`...

> Ah yes, thank you for reminding me. A good future-proof name would take into account the most likely extensions of this. In [#780](https://github.com/Colvars/colvars/pull/780) you mentioned "optimal alignment, RMSD, COM...

I wonder if this bug is related to https://github.com/Colvars/colvars/pull/808.

Is it possible to use multiple grids in a bias? I am not sure if the configuration syntax in PR #710 implies that there is only one grid allowed.

I guess the following functions can be shared between `colvarproxy_impl` in `colvarproxy_cudaglobalmaster.C` and `colvarproxy_namd`: ``` load_atoms_pdb load_coords_pdb update_target_temperature log error check_atom_id update_atom_properties output_stream flush_output_stream flush_output_streams close_output_stream close_output_streams backup_file set_unit_system ```

> Current state is that the two classes are now separated, and things seem to be working. > > Next I'll be taking a look at consolidating with the `CUDAGlobalMaster`-based...

It seems with this PR and NAMD's main branch, the interface volmap tests are still failed either in the CPU build or in the GPU build. Do I miss something?