Regarding that benchmark... so blosc2 and lz4 compression? Both can benefit from SIMD. I wonder how well optimized those ARM versions are they're using. Do the

Regarding that benchmark... so blosc2 and lz4 compression? Both can benefit from SIMD. I wonder how well optimized those ARM versions are they're using. Do they even use NEON? Pretty likely x86 versions use SSE2 or AVX2. So is that really a CPU performance benchmark or blosc2 and LZ4 library implementation quality benchmark?