Intel® oneAPI Math Kernel Library
Ask questions and share information with other developers who use Intel® Math Kernel Library.
Announcements
FPGA community forums and blogs on community.intel.com are migrating to the new Altera Community and are read-only. For urgent support needs during this transition, please visit the FPGA Design Resources page or contact an Altera Authorized Distributor.
7234 Discussions

zgemm3m and MIC automatic off load

AndrewC
New Contributor III
‎08-03-2016 09:56 AM
510 Views

I have noted that zgemm will be taken by AO to the Xeon PHI, but not zgemm3m. This is a bit annoying as AO can give a significant boost to zgemm performance, but zgemm3m is faster than ZGEMM normally.

For example on my 8 core machine, ZGEMM3M takes 46s, ZGEMM 62s for a 8192 x 8192 matrix

With AO to the Phi, ZGEMM takes 12.5s, but ZGEMM3M is not accelerated by AO at all.

 

 

 

0 Kudos

Link Copied

1 Reply
AndrewC
New Contributor III
‎08-03-2016 01:07 PM
510 Views

Of course I suppose one could switch from zgemm3m to zgemm if I could detect if AO was enabled, but there does not seem to be an API for that. mkl_mic_enable(),.mkl_mic_disable() exist, but no mkl_mic_isenabled() ?

0 Kudos
Copy link
Reply

Community support is provided Monday to Friday. Other contact methods are available here.

Intel does not verify all solutions, including but not limited to any file transfers that may appear in this community. Accordingly, Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade.

For more complete information about compiler optimizations, see our Optimization Notice.