They were temporarily disabled during testing of the normal kernels, but should be straightforward to turn on again.
- Status changed from New to In Progress
- Assignee set to Erik Lindahl
- Target version changed from 4.6 to 4.6.1
- Target version deleted (
- Affected version set to 4.6
- Status changed from In Progress to Resolved
we can close this one also
I don't think the underlying issue is resolved. My patch just stopped wrong results being generated. Resolution ought to be removing or fixing the broken code. This might be trickier now with PD (and perhaps simple search?) removed.
- Status changed from Resolved to In Progress
- Tracker changed from Bug to Feature
Changing to a feature; we should still improve this for post 5.0 to improve performance, but Mark's fix means we no longer produce any incorrect results.
- Related to Task #1971: Removing buggy features vs. keeping workflows added
Actually I think the gmx_incons I put into do_nonbonded() back then has meant that this code path hasn't worked at all since 4.6.x.
While we plan to add back some all-vs-all support to the Verlet scheme, it seems like the group-scheme implementation may as well be removed ASAP.
Also available in: Atom