Skip to content

Mark compilation-unit-internal symbols static - #9773

Draft
akx wants to merge 2 commits into
python-pillow:mainfrom
akx:statique
Draft

Mark compilation-unit-internal symbols static#9773
akx wants to merge 2 commits into
python-pillow:mainfrom
akx:statique

Conversation

@akx

@akx akx commented Jul 6, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

This should allow compilers to better inline those functions.

Along the way, this:

  • deletes ImageAccessDelete (no-op since 2009)
  • deletes debug_clip_tree (added in 9a9d3a0, 2020). Since it wasn't used outside of Draw.c, it would've been marked static, and at that point become entirely dead code. In addition, it would not have compiled in its current form anyway, since fprintf() and friends aren't around

@codspeed

codspeed Bot commented Jul 6, 2026 •
edited
Loading

Copy link
Copy Markdown

Merging this PR will degrade performance by 9.46%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 4 improved benchmarks
❌ 13 regressed benchmarks
✅ 623 untouched benchmarks
⏩ 338 skipped benchmarks1

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

Benchmark BASE HEAD Efficiency
❌ test_scale[1237x811-L-0.8-NEAREST] 5 ms 6.3 ms -20.65%
❌ test_scale[1237x811-L-2.14-NEAREST] 39.7 ms 49.7 ms -20.16%
❌ test_transform[1237x811-L-NEAREST-AFFINE] 16.7 ms 20.9 ms -20.07%
❌ test_scale[1237x811-LA-2.14-NEAREST] 84.6 ms 101.9 ms -16.97%
❌ test_scale[1237x811-I-2.14-NEAREST] 84.7 ms 101.9 ms -16.89%
❌ test_scale[1237x811-F-2.14-NEAREST] 84.6 ms 101.8 ms -16.87%
❌ test_scale[1237x811-RGBA-2.14-NEAREST] 84.7 ms 101.8 ms -16.85%
❌ test_scale[1237x811-RGB-2.14-NEAREST] 84.7 ms 101.8 ms -16.81%
❌ test_scale[1237x811-F-0.8-NEAREST] 14.1 ms 16.5 ms -14.59%
❌ test_scale[1237x811-I-0.8-NEAREST] 14.1 ms 16.5 ms -14.59%
❌ test_scale[1237x811-RGB-0.8-NEAREST] 14.1 ms 16.5 ms -14.58%
❌ test_scale[1237x811-LA-0.8-NEAREST] 14.1 ms 16.5 ms -14.58%
❌ test_scale[1237x811-RGBA-0.8-NEAREST] 14.1 ms 16.5 ms -14.58%
⚡ test_transform[1237x811-L-BILINEAR-AFFINE] 107.1 ms 81.7 ms +30.95%
⚡ test_transform[1237x811-L-BICUBIC-AFFINE] 172.2 ms 146.9 ms +17.22%
⚡ test_transform[1237x811-LA-BILINEAR-AFFINE] 178 ms 152.9 ms +16.38%
⚡ test_transform[1237x811-RGB-BILINEAR-AFFINE] 217.3 ms 192 ms +13.2%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

Comparing akx:statique (cc42e3a) with main (38c17ed)

Open in CodSpeed

Footnotes

  1. 338 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

@akx

akx commented Jul 6, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor Author

That's a surprising result from CodSpeed!

Probably has to do with the different CPUs: *AMD EPYC 7763 64-Core Processor* → *AMD EPYC 9V74 80-Core Processor*

Oh -- #9760 should be merged first before we take another look at this, because that one does significant things to Image.filter()!

@akx
akx force-pushed the statique branch 2 times, most recently from cda506f to 7e0524d Compare July 6, 2026 13:48
@akx akx mentioned this pull request Jul 29, 2026
Merged
@radarhere

radarhere commented Sep 10, 2026

Copy link
Copy Markdown
Member

#9760 has been merged.

@akx
akx force-pushed the statique branch 4 times, most recently from 02d1911 to 0fb8527 Compare September 16, 2026 10:51
@akx

akx commented Sep 18, 2026

Copy link
Copy Markdown
Contributor Author

I don't really have an idea why this would regress those 3 NEAREST benchmarks aside from measurement fluke or some codegen esoterica...

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@akx @radarhere