... 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 tofloat64
)
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, ...)