oblas

blas-like routines to solve systems in finite fields [gf2, gf4, gf16 and gf256]

Optimizing for different archs

NEON: make CPPFLAGS+="-DOBLAS_NEON"
SSE: make CPPFLAGS+="-DOBLAS_SSE"
AVX: make CPPFLAGS+="-DOBLAS_AVX -DOCTMAT_ALIGN=32"
AVX512: make CPPFLAGS+="-DOBLAS_AVX -DOCTMAT_ALIGN=64"

Customizing

Edit tablegen.c to change polynomials.

TODO

AES-NI support
common multibyte fields (GF2_16 etc)