「view this page in B3 βῆτα server」

Revisions №62002

branch: master 「№62002」
Commited by: Rocco Moretti
GitHub commit link: 「28b344ef7da98efe」 「№6353」
Difference from previous tested commit:  code diff
Commit date: 2023-04-25 12:12:13
release.PyRosetta.MinSizeRel release.PyRosetta.Release scientific.ligand_scoring_ranking scientific.mp_f19_sequence_recovery

Merge pull request #6353 from RosettaCommons/roccomoretti/speed_VDDistanceMatrix Speed up NBR distance calculation. It looks like the slow bit of loading types from the Chemical Components Dictionary is the find_nbr_dist() calc. The slow bit looks to be data structure access in the Floyd-Warshall. Right now we're using a map of maps. If we can use compact data storage (and integer indexes), we can speed this up greatly. The find_nbr_dist() function gets ~50 times faster, though given the total time we spend in it, the overall speedup due to this fix is probably ~1% at most (depending on how many CCD components you have.)

...
Test: mac.clang.python311.release.PyRosetta.MinSizeRel

 View log

Loading...

 View log in dialog  View log in log in separate window
Test: mac.clang.python311.release.PyRosetta.Release

 View log

Loading...

 View log in dialog  View log in log in separate window
Test: ubuntu-20.04.clang.python36.scientific.mp_f19_sequence_recovery

 View log

Loading...

 View log in dialog  View log in log in separate window