 Alexandr Guzhva
|
d0ba4c04ca
|
Faster versions of fvec_op_ny_Dx for AVX2 (#2811)
Summary:
Pull Request resolved: https://github.com/facebookresearch/faiss/pull/2811
Use transpose_AxB kernels to speedup computations.
Reviewed By: mdouze
Differential Revision: D44726814
fbshipit-source-id: a1dd3e289f4ed564a5bece699bee0af88c9925b0
|
2023-04-24 15:03:45 -07:00 |
|