gm

... high performance float32 matrix and vector math library using SIMD for graphics software.

use SIMD instructions (at least where it makes sense)
mutate matrices/vectors in place
use float32 math (instead of converting values to float64)

Documentation | Benchmarks

Status:

Feature set is incomplete. I am implementing methods once I need them. PRs are welcome (also, feel free to kindly ask for missing functions).
API may change.
ARM falls back to pure go implementations, because of golang/go#7300
Any feedback is welcome (regarding performance, implementations, API, ...)