Conversation
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## develop #1357 +/- ##
===========================================
- Coverage 90.86% 90.84% -0.03%
===========================================
Files 135 135
Lines 14683 14658 -25
===========================================
- Hits 13342 13316 -26
- Misses 1341 1342 +1 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
garrettwrong
left a comment
There was a problem hiding this comment.
Thanks. Try to hit the non cu code stuff and let me know when that's ready, then I'll try a pass on the cu code.
|
@garrettwrong I hit all the non-cu stuff. Marking for re-review. |
|
I think this is okay to continue on ... Hopefully you can merge it in soon and then should mainly just be any docstring cleanup and the long form documentation you've been working on to wrap up the reorg. |
Ok, great. Thanks @garrettwrong. I'll update the branch and move it along. |
The main focus of this PR is adopting the optimized J-sync code for all CL algos that can use it. This involved swapping the optimized code in for the pre-existing code in the
JSyncmodule, validating that outputs were not altered for adopting algos (up to multiplication by -1 on the signs vector), and ensuring that gpu dispatching is working as expected.Additionally, I ran some experimental recons of 10081 and updated the gallery experiment to be reflective of the recon in the associated publication. I will run the gallery experiment after initial reviews. I also made some optimizations (~4X-5X speedup) to the
_estimate_inplane_rotationsmethod which is currently one of the bottlenecks of the Cn algos.