Re: [ublas] Matrix multiplication performance

Subject: Re: [ublas] Matrix multiplication performance
From: Nasos Iliopoulos (nasos_i_at_[hidden])
Date: 2016-01-28 11:54:14

You should be able to do it if you want to convert a sparse to a dense matrix like 

.. matrix<double>( my_sparse_matrix) ..

we just don't want surprise implicit conversions.

-N

Date: Thu, 28 Jan 2016 02:09:10 +0100
From: jdurancomas_at_[hidden]
To: ublas_at_[hidden]
Subject: Re: [ublas] Matrix multiplication performance

Referring to the point 1:
1. It is called  only if both arguments are dense matrices, or one dense matrix and one dense vector. We should disable any implicit conversions to avoid surprises like converting a sparse matrix to a dense one

I think that it should be possible uncompress partially spare matrices in small buffers, use SIMD for calculations and then assign the result to result matrix.
Joaquim Duran

2016-01-27 16:18 GMT+01:00 nasos <nasos_i_at_[hidden]>:

_______________________________________________

_______________________________________________
ublas mailing list
ublas_at_[hidden]
http://lists.boost.org/mailman/listinfo.cgi/ublas
Sent to: nasos_i_at_[hidden]