MKL_DIRECT_CALL works great!

MKL_DIRECT_CALL works great!

I Just wanted to thank the MKL team for MKL_DIRECT_CALL. It works well with small matrices as well as large matrices.  I am seeing a 2x performance improvement when the matrices are small. Previously I had my own _inline gemm for small matrices as calling MKL zgemm was inefficient, but the new feature has allowed me to clean up my code as well as speed it up....

2 posts / 0 new
Last post
For more complete information about compiler optimizations, see our Optimization Notice.

thanks Andrew. 

Leave a Comment

Please sign in to add a comment. Not a member? Join today