Consider merging this with SIMD Everywhere project

[mr-c]

https://github.com/simd-everywhere/simde/

I think we have wider coverage of x86 intrinsics, but maybe you have some optimized implementations we lack.

https://github.com/simd-everywhere/implementation-status/blob/main/x86.md

There are a total of 6876 SIMD functions on x86, 2937 (42.71%) of which have been implemented in SIMDe so far.

Families that are 100% implemented

• AVX2
• AVX512VPOPCNTDQ
• AVX512_4VNNIW
• AVX512_BITALG
• AVX512_VBMI
• AVX512_VNNI
• AVX512_VP2INTERSECT
• CRC32
• F16C
• FMA
• GFNI
• MMX
• PCLMULQDQ
• SSE2
• SSE3
• SSE4.1
• SSSE3
• VPCLMULQDQ

Partial:

• AES: Implemented: 5 of 6 (83.33%)
• AVX: Implemented: 296 of 298 (99.33%)
• AVX512BW: Implemented: 337 of 790 (42.66%)
• AVX512CD: Implemented: 21 of 42 (50.00%)
• AVX512DQ: Implemented: 112 of 376 (29.79%)
• AVX512F: Implemented: 1087 of 2824 (38.49%)
• AVX512_BF16: Implemented: 9 of 38 (23.68%)
• AVX512_FP16: Implemented: 17 of 1105 (1.54%)
• AVX512_VBMI2: Implemented: 3 of 150 (2.00%)
• AVX_VNNI: Implemented: 8 of 16 (50.00%)
• SSE: Implemented: 239 of 241 (99.17%)
• SSE4.2: Implemented: 9 of 19 (47.37%)