Skip to content

WIP: Integrate support for GlobalMaster and CUDAGlobalMaster#914

Draft
giacomofiorin wants to merge 4 commits into
masterfrom
integrate_cudagm
Draft

WIP: Integrate support for GlobalMaster and CUDAGlobalMaster#914
giacomofiorin wants to merge 4 commits into
masterfrom
integrate_cudagm

Conversation

@giacomofiorin
Copy link
Copy Markdown
Member

Currently, GlobalMaster (GM) is a required component of the default interface between NAMD and Colvars; this class interacts with the class colvarproxy_namd to communicate all data related to atoms, groups and unit cell.

PR #783 introduces an alternative interface that is much faster in GPU-resident runs, but does not support the other use cases (pure CPU or GPU-offload) plus some specialized features of GlobalMaster. Additionally, the CUDAGlobalMaster (CUDAGM) client introduced in that PR uses a different class to interact with Colvars, called colvarproxy_impl, which duplicates some but not all of the code of colvarproxy_namd.

This PR, currently still in progress, aims to unify the colvarproxy_impl and colvarproxy_namd classes, so that the end product (still called colvarproxy_namd for consistency) can handle both GM and CUDAGM interfaces.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant