Build: #2 was successful Changes by Srikrishna Sekhar

Stages & jobs

  1. Default Stage

  2. Test

Build result summary

Details

Completed
Queue duration
< 1 second
Duration
46 minutes
Labels
None
Revision
2cf4071aa63ff380efca1337793b13f32fe52d11
Total tests
696
Successful since
#1 ()

Tests

Code commits

Author Commit Message Commit date
Srikrishna Sekhar Srikrishna Sekhar 2cf4071aa63ff380efca1337793b13f32fe52d11 CAS-14527 : Improve AWP2 memory performance
The wtcf for A-term was duplicated across all W planes. This is not
strictly necessary, since A term wtcf is independent of W. The code has
been refactored to reflect this.

From VLASS testing this was a ~ 20-30% drop in sustained memory load.
Srikrishna Sekhar Srikrishna Sekhar b411209da523195179f1b816ae13358f23332808 CAS-14527 : Memory improvement to W-plane CFs
Instead of allocating a large buffer and trimming down to max W support
size, calculate the max W support size a priori and create a buffer that
is 2X that size (same as the post-trim size before this change).

From VLASS testing, this yields a ~ 40% drop in peak memory required,
although this is very dependent on max W support size.
Srikrishna Sekhar Srikrishna Sekhar af3d7bb816351e0c35d926a653c7d21e660f9c24 CAS-14527 : Incremental gridding performance
The gridder discarded (prior to this commit) the per-W support size and
always used the largest support size.

This change respects the computed per-W support size during gridding,
therefore small W planes see a speedup and do not waste time gridding
with effectively padding.

Jira issues

IssueDescriptionStatus
Unknown Issue TypeCAS-14527Could not obtain issue details from Jira

Shared artifacts

Artifact File size
ManyLinux228 Casatestutils 157 KB
ManyLinux228 Python 3.12 Tool wheel 75 MB