vanilla C and/or arm64 support?

[josharian]

I am trying to get serenade.ai to execute natively on an m1 mac (not rosetta). This is one of a very small number of dependencies that cannot be compiled at all on arm64. (Unless I'm missing something, which would be great.)

This is not on a critical performance path for serenade, so a vanilla C implementation (with appropriate build flags) would suffice.

I am also personally familiar enough with arm64 SIMD instructions that I could probably cobble together a reasonably optimized version, if needed.

Would such PRs be welcome?

(I would anticipate a little bit of back and forth; my C is rusty. Or actually, it's probably Go-y. :P)

[kpu]

Such PRs would be would be welcome. For ARM, we've been using https://github.com/google/ruy though the lack of postprocessing https://github.com/google/ruy/issues/307 is cramping my style (and requiring an extra round-trip to RAM to add bias terms).