search

google / gemmlowp

Low-precision matrix multiplication
Apache License 2.0
1.77k stars 451 forks source link

Add 16-bit support for AVX2 #201

Closed talumbau closed 4 years ago

talumbau commented 4 years ago

Equivalent to 16-bit SSE support added in:

https://github.com/google/gemmlowp/commit/76de7f790a2d65b88a06a2fbe9cfb5e7a70aebb6#diff-a414deff691f120bc86c429de8e68319