OMP threading

OMP threading


We are using only a selected few routines from Intel MKL such as dgemm, daxpy, ddot, dgemv, etc. For all the other blas/lapack routines we compile and link in the actual lapack/blas routines. My questions are as follows:

(i) Is dgemm, daxpy, ddot or dgemv parallelized ?
(ii) Do I need to use the full suite of Intel MKL routines to take advantage of parallelism in a select few routines?

By the way, we are using Intel MKL 7.2


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

Why use MKL for some functions, but not for others? I'm curious.

DGEMM is certainly parallized. The other functions from Levels 1 and 2 of the BLAS are not parallelized.


Leave a Comment

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