Further tidy up codegen and somewhat rework the API #149
+2,577
−4,519
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
We exposed an eclectic set of inherent methods on each SIMD type. Most of them have now been subsumed by trait-based operations, so we should get rid of them to avoid redundancy.
Since the intent is to move from inherent methods to trait methods, it's important to provide a way to import all the traits so the methods are in scope, without having to import them all separately or wildcard-import the entire library.
valadaptive
force-pushed
the
api-cleanups
branch
from
395271d to
de443b6
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Depends on #148.
This PR intertwines codegen and API changes, unfortunately. Some of the codegen cleanups depend on it.
All the inherent impl methods on individual SIMD types are now gone entirely. Most of them were previously already accessible via the
core::opstraits,SimdBase/SimdInt/SimdFloat/SimdMask, or individual bespoke traits. The exceptions were split/combine. There was no inherentsplitmethod at all, and I've removed the inherentcombinemethod. In its place I've added newSimdCombineandSimdSplittraits, which are implemented for types of the proper vector width.Since all the operations need to be accessed via trait methods, I've added a
preludemodule tofearless_simd, which re-exports just the traits. This lets API consumers import every necessary trait to access all the methods, without being forced to either import every trait individually or just wildcard-import every single item fromfearless_simd.I'm not entirely happy with the current
SimdCvtTruncateandSimdCvtFloattraits--it's unclear which is the source and which is the destination,to_intandto_floatare technically implemented for allSimdFloatandSimdInttypes even though they only work on 32-bit ones, and they require passing in the full vector equivalent (withStype parameter) as a type. Improving that is future work.Aside from all that, this PR is about further centralizing the codegen in preparation for adding documentation generation. I want to make it as operation-centric as possible, moving op-specific special cases out of each module generation method and into the operation metadata itself.