Tracking Issue for `stdarch_x86_avx512_vl_u32_to_f32_conversions`

[Shnatsel]

Feature gate: #![feature(stdarch_x86_avx512_vl_u32_to_f32_conversions)]

This is a tracking issue for the AVX-512VL intrinsics that convert packed unsigned 32-bit integers to single-precision floating-point values.

These intrinsics correspond to the VCVTUDQ2PS instruction. They are documented in the Intel Software Developer’s Manual, but are not listed in the Intel Intrinsics Guide.

Public API

#[target_feature(enable = "avx512f,avx512vl")]
pub const fn _mm_cvtepu32_ps(a: __m128i) -> __m128 { /* ... */ }

#[target_feature(enable = "avx512f,avx512vl")]
pub const fn _mm_mask_cvtepu32_ps(
    src: __m128,
    k: __mmask8,
    a: __m128i,
) -> __m128 { /* ... */ }

#[target_feature(enable = "avx512f,avx512vl")]
pub const fn _mm_maskz_cvtepu32_ps(
    k: __mmask8,
    a: __m128i,
) -> __m128 { /* ... */ }

#[target_feature(enable = "avx512f,avx512vl")]
pub const fn _mm256_cvtepu32_ps(a: __m256i) -> __m256 { /* ... */ }

#[target_feature(enable = "avx512f,avx512vl")]
pub const fn _mm256_mask_cvtepu32_ps(
    src: __m256,
    k: __mmask8,
    a: __m256i,
) -> __m256 { /* ... */ }

#[target_feature(enable = "avx512f,avx512vl")]
pub const fn _mm256_maskz_cvtepu32_ps(
    k: __mmask8,
    a: __m256i,
) -> __m256 { /* ... */ }

Steps / History

(Remember to update the S-tracking-* label when checking boxes.)

• Implementation Add missing avx512vl intrinsics for u32->f32 conversions stdarch#2197
• Final comment period (FCP)¹
• Stabilization PR

Unresolved Questions

None at this time.

Footnotes

1. https://std-dev-guide.rust-lang.org/feature-lifecycle/stabilization.html ↩