FastWalkthrough (CPU version)
The fast version of the Walkthrough algorithm, currently only running on CPU. Functionality is already here in this draft, I just need to remove some of the timing code used for the PUB note.
Merge request reports
Activity
- Resolved by Daniel Thomas Murnane
added 2 commits
added 1 commit
- 2e6a7568 - Revert to dev version of reader files and cc_and_walk files
added 21 commits
-
6c493d88...63336250 - 20 commits from branch
dev
- 1da74ee2 - Merge in feature name fix from dev
-
6c493d88...63336250 - 20 commits from branch
@scaillou Okay, this is fixed. Should be good to go.
- Tested with Rel24 scored graphs, runs in less than 1 second per graph (most of this is data loading and saving)
- Resolved by Daniel Thomas Murnane
This is great! Would you please to add an exemplar configuration in at least one Example in the example folder so that users know how to use it properly.
- Resolved by Daniel Thomas Murnane
- Resolved by Daniel Thomas Murnane
Hi @dmurnane I have added a few fixes to ensure compatibility with the new naming scheme and the eval. Can you have a quick look at it ?
- Tested with CTD23 scored graphs (infer + eval) with the old naming scheme
- Tested with CTD23 scored graphs (infer + eval) with the new naming scheme
Edited by Sylvain Caillou
added 3 commits
-
9dd88a81...bfea733c - 2 commits from branch
dev
- fc754bb2 - Merge branch 'dev' of ssh://gitlab.cern.ch:7999/gnn4itkteam/acorn into dmurnane_faster_walkthrough
-
9dd88a81...bfea733c - 2 commits from branch
Please register or sign in to reply