Parallelize MLPs better

accosmin / nano

C++ library [machine learning & numerical optimization] - superseeded by libnano

MIT License

1 stars 0 forks source link

Closed accosmin closed 8 years ago

accosmin commented 8 years ago

Currently the convolution networks scale much better with the number of cores (~linearly) than MLPs (~threads/2).

accosmin commented 8 years ago

It works OK in the end for a larger number of samples.