Tracking issue for portable-simd migration

[Nugine]

Current supported instruction sets (2022-11-18):

• AVX2
• SSE4.1
• SSSE3
• SSE2
• ARM NEON
• AArch64 NEON
• WASM SIMD128

Upstream blockers: (2022-11-18)

• Bad codegen for bitwise OR/AND masks rust-lang/portable-simd#303
• Suboptimal codegen for mask8x32 all rust-lang/portable-simd#316

[gilescope]

That would be nice - would simd wasm come out of that as a side effect?

[Nugine]

This repo supports x86, arm and wasm by hand-written abstractions now.

Supported instruction sets:

• SSE4.1
• AVX2
• WASM SIMD128
• NEON (arm, unstable, nightly)
• NEON (aarch64, unstable, nightly)

portable-simd will simplify the code. It may not affect the performance as the current implementations are highly optimized.

[Nugine]

Upstream blockers:

• Bad codegen for bitwise OR/AND masks rust-lang/portable-simd#303
• Suboptimal codegen for mask8x32 all rust-lang/portable-simd#316