As discussed in #18, most of the code in simd_check.c is intel specific. Furthermore, it seems like the interface used in simd_instruction.h, which defaults to sse4.1 (for met at least) doesn't work with clang, making it difficult for users of arm Macs.
This PR addresses both these issues with hacks that use AVX2 for everything when on arm. But since this is the AVX2 interface into simde, it should still be portable (on arm it will call the appropriate neon calls).
As discussed in #18, most of the code in simd_check.c is intel specific. Furthermore, it seems like the interface used in simd_instruction.h, which defaults to sse4.1 (for met at least) doesn't work with clang, making it difficult for users of arm Macs.
This PR addresses both these issues with hacks that use AVX2 for everything when on arm. But since this is the AVX2 interface into simde, it should still be portable (on arm it will call the appropriate neon calls).
This builds and runs simple examples on