From: julie langou
Date: Fri, 19 Nov 2010 11:57:36 -0700
the current work for higher precision is available since LAPACK 3.2.0.
Here is the release notes: http://www.netlib.org/lapack/lapack-3.2.html
The XBLAS project is here: http://www.netlib.org/xblas/index.html
As you will see generalized diagonalization has not been implemented and it is
not planned anytime soon.
Another way to do is just to use the ability of the Intel Compiler that can
work with quads (REAL*16)
This would require to modify all the LAPACK routines you need by hand,
replacing double precision by real*16.
This would work.
On Nov 7, 2010, at 9:30 PM, Chen Liu wrote:
Dear Lapack developer,
Hello! I am using the Lapack right now. It is very helpful for my
research. I appreciate it!
Right now, I am doing some project requiring accurater linear algebra
tools: generalized diagonalization. I want to do that for quadruple
precision. So I am wondering, how I could use Lapack for higher precision?
Thank you very much for your kindness and help!
Department of Physics
590 Commonwealth Avenue, Room 315
Boston, MA 02215
Lapack mailing list