feat: Enable SIMD optimizations

[matthewfeickert]

• Add x86_64-microarch-level as a 'build' requirement.
  • microarch_level 4 not supported yet so only add level 1 to 3.
  • c.f. Easier cross-compiling for level 4? microarch-level-feedstock#5
• Set the build number based on the microarch_level.
• Add conda_build_config.yaml with microarch_level.
• c.f. https://prefix.dev/blog/building_cpu_optimized_packages
• Bump build number.

Checklist

• Used a personal fork of the feedstock to propose changes
• Bumped the build number (if the version is unchanged)
• [N/A] Reset the build number to 0 (if the version changed)
• Re-rendered with the latest conda-smithy (Use the phrase @conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
• Ensured the license file is being packaged.

[matthewfeickert]

@conda-forge-admin, please rerender

[conda-forge-webservices]

Hi! This is the friendly automated conda-forge-linting service.

I wanted to let you know that I linted all conda-recipes in your PR (recipe/meta.yaml) and found some lint.

Here's what I've got...

For recipe/meta.yaml:

• Selectors are suggested to take a <two spaces>#<one space>[<expression>] form. See lines [29]

[conda-forge-webservices]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe/meta.yaml) and found it was in an excellent condition.

[matthewfeickert]

@conda-forge/iminuit this is ready for review. Let me know if you have any questions. 👍

[henryiii]

Do you know if PR's are actually making anything faster? Some code bases can effectively use vector instructions and get a nice 2x+ speed up and some don't see much benefit.

[matthewfeickert]

Do you know if PR's are actually making anything faster? Some code bases can effectively use vector instructions and get a nice 2x+ speed up and some don't see much benefit.

@henryiii Good question. I don't have any explict checks to show at the moment. I'll defer to you if you think this is worth doing here (and for the other ones that I've opened up), and if so I'll do some custom builds and test comparisons when I'm back from work travel in a few weeks.

[matthewfeickert]

Placed into draft given that needs to be rebased following PR #108 and needs conda-forge/microarch-level-feedstock#10 to be resolved before can be merged.

[HDembinski]

I cannot review this, but I appreciate the effort and I think it sound all good.

[HDembinski]

That being said, iminuit will not profit from such optimizations, because the minimizing algorithm is typically not the bottleneck, it is the calculation of the cost function, which ideally should be JIT-compiled for the specific architecture.

[matthewfeickert]

That being said, iminuit will not profit from such optimizations, because the minimizing algorithm is typically not the bottleneck

Sounds good. No need to add more build complexity if it won't help, so closing this. 👍 Thanks, Hans, for the info!