Open discussion for MAGMA library (Matrix Algebra on GPU and Multicore Architectures)
Do all the functions of MAGMA on fermi GPU outperform those in MKL on CPU xeon 5600 series with flags "openmp" and "parallel" turned on?
In the presentation slides, there is only one slide showing MAGMA is faster than MKL for LU decomposition. I don't know what about the other operations, such as dheevd?
- Posts: 3
- Joined: Fri May 20, 2011 11:23 am
I don't know especially for this serie but usually yes, except that the multiplication factor will change from one function to another.
- Posts: 41
- Joined: Tue Mar 08, 2011 12:38 pm
Return to User discussion
Who is online
Users browsing this forum: Baidu [Spider], Bing [Bot], Google [Bot] and 1 guest