Thread 'Using Intel Math Kernel Library and Intel Xeon Phis with BOINC'

Message boards : API : Using Intel Math Kernel Library and Intel Xeon Phis with BOINC
Message board moderation

To post messages, you must log in.

AuthorMessage
Andrew Ekstrom

Send message
Joined: 28 Jul 15
Posts: 1
United States
Message 63269 - Posted: 28 Jul 2015, 19:11:57 UTC

I have a home computer than has a pair of Xeon Phi coprocessors. These coprocessors work in a similar fashion to a GPU but, are easier to program for. My Phis have a total of 456 processing threads that can be used separately or in parallel by BOINC. If someone uses Intel's Math Kernel Library, there is an automatic offload option for these Phi's, which means there is no required extra programming at all. I'm wondering if it is possible to use the MKL with BOINC. If so, I'd be able to run through large data simulations in little to no time at all.
ID: 63269 · Report as offensive
Coleslaw
Avatar

Send message
Joined: 23 Feb 12
Posts: 198
United States
Message 63272 - Posted: 28 Jul 2015, 20:39:19 UTC - in response to Message 63269.  

The problem is getting a project that is willing to support it. My last efforts in finding one resulted in:
1. Tried them and they weren't efficient enough
2. Not enough people with them to bother
3. Don't have the resources to bother even if we wanted too....
ID: 63272 · Report as offensive

Message boards : API : Using Intel Math Kernel Library and Intel Xeon Phis with BOINC

Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.