Skip to content

[wip] Imageitem performance#669

Closed
campagnola wants to merge 12 commits intopyqtgraph:masterfrom
campagnola:imageitem-performance
Closed

[wip] Imageitem performance#669
campagnola wants to merge 12 commits intopyqtgraph:masterfrom
campagnola:imageitem-performance

Conversation

@campagnola
Copy link
Copy Markdown
Member

Working on improving update performance of ImageItem:

  • ensure that ubyte data can be passed through to QPixmap without any extra overhead
  • optimize levels / lookup tables / clipping / channel ordering where possible (maybe we can use numba to combine multiple steps into a single loop)

Fixes #667

@campagnola campagnola force-pushed the imageitem-performance branch from 59a5eff to 4556fb6 Compare July 3, 2018 05:42
@j9ac9k j9ac9k changed the base branch from develop to master October 28, 2020 17:00
@j9ac9k
Copy link
Copy Markdown
Member

j9ac9k commented Oct 28, 2020

This PR was still pointing to develop, not master... hey look at that, it's no longer 200+ commits in a PR :D

@joshdoe
Copy link
Copy Markdown

joshdoe commented Nov 25, 2020

I tested out this PR in my app, and for a 4k30hz video the display framerate goes from 2.5Hz up to 14Hz, quite an improvement.

@j9ac9k
Copy link
Copy Markdown
Member

j9ac9k commented Jan 5, 2021

Closing/Opening to trigger run in new CI

@j9ac9k j9ac9k closed this Jan 5, 2021
@j9ac9k j9ac9k reopened this Jan 5, 2021
@outofculture
Copy link
Copy Markdown
Contributor

Okay, so considering #1466 is about to get merged, I dug in and figured out what other improvements are in this branch. The primary optimization here was pre-allocating the processing buffer, which is included as part of #1466. This branch does fix a bug (#667), improves docstrings, profiler and code clarity in a few places, and adds an Air-Speed-Velocity-based benchmark tool. These two branches have conflicting changes, so I made a new branch with those ironed out, and a few other things cleaned up. The hopeful list at the beginning of this PR is, as yet, undone, so I'm going to recommend we close this PR for now and make a new one out of the merged branches that better describes its functional intent.

@j9ac9k
Copy link
Copy Markdown
Member

j9ac9k commented Jan 20, 2021

This PR got super seeded by #1501 and #1466 thanks for the PR @campagnola !

@j9ac9k j9ac9k closed this Jan 20, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

ndarray.dtype check fails erroneously leading to performance degradation

4 participants