Tracking Issue for AVX512_FP16 intrinsics

[sayantn]

Feature gate: #[feature(stdarch_x86_avx512_f16)]

This is the tracking issue for the AVX512_FP16 (and related AVX_NE_CONVERT) intrinsics. This depends on the following features

• Tracking Issue for AVX512 intrinsics #111137
• Tracking Issue for f16 and f128 float types #116909

Public API

This feature covers all of the intrinsics from the avx512fp16 feature and some remaining intrinsics from the avxneconvert feature (more specifically, the __m128h, __m256h and __m512h types, and all intrinsics that use those types or have a f16 parameter)

Implementation History

• Implement AVX512_FP16 stdarch#1605

Steps

• Final Comment Period (FCP)
• Stabilization PR

Unresolved Questions

None yet

[sayantn]

@Amanieu @tgross35 can we start stabilizing some of these intrinsics? I think stabilization of the types + the intrinsics that don't directly use f16 is okay (after #111137 obviously, but we can start the discussion)

[tgross35]

For anything that doesn't have f16 in the function signature, I think it should be fine to move those forward. Anyone who needs to work with the types on stable can use half.

[Amanieu]

Proposing stabilization of intrinsics that use f16xN vector types, but not intrinsics that use f16 directly since that is still blocked on stabilization of f16 (#116909 ).

@rfcbot merge

[rfcbot]

Team member @Amanieu has proposed to merge this. The next step is review by the rest of the tagged team members:

• @Amanieu
• @BurntSushi
• @dtolnay
• @joshtriplett
• @m-ou-se

No concerns currently listed.

Once a majority of reviewers approve (and at most 2 approvals are outstanding), this will enter its final comment period. If you spot a major issue that hasn't been raised at any point in this process, please speak up!

See this document for info about what commands tagged team members can give me.