You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
As very small vectors and matrices are used on testing, when using AVX2 backend the operations are not really tested as to use those instructions sufficient elements need to be on the vectors or matrices.
The text was updated successfully, but these errors were encountered:
With this pull request the implementation of the AVX2 backend should be ready and pretty stable. Apart from that, now all fors that do calculations advance by LIBLINALG_PARALLEL_FLOATS variable instead of by the hardcoded value. Making it more consistent. This pull request fixes issues #20, #17 and #23.
As very small vectors and matrices are used on testing, when using AVX2 backend the operations are not really tested as to use those instructions sufficient elements need to be on the vectors or matrices.
The text was updated successfully, but these errors were encountered: