Improve Intel MKL Performance for Small Problems: The Use of MKL_DIRECT_CALL

One of the big new features introduced in the Intel MKL 11.2 is the greatly improved performance for small problem sizes.

Authored by Zhang, Zhang (Intel) Last updated on 06/07/2017 - 10:26

The Alternatives for Intel® IPP Legacy Small Matrices Domain


Authored by Igor Astakhov (Intel) Last updated on 06/01/2017 - 11:27

Tips to measure the performance of Intel® MKL with small matrix sizes

The time required by the first Intel® MKL call should be ignored for the performance measurements. The first Intel MKL call has overhead due to buffer allocation and thread initialization. Ignoring the first Intel MKL call gives more consistent times for small problems.
Authored by Ying H. (Intel) Last updated on 06/07/2017 - 12:08
For more complete information about compiler optimizations, see our Optimization Notice.