Hi,
is there performant code for a dense and sparse LU factorization
with partial pivoting? Concerning the sparse part, fill-in reducing
steps are not needed.
thanks for any hint.
if you want to use LU for solving systems of equations, the xTRSM part of the clAmdBlas library may be of some use for you.
http://developer.amd.com/gpu/appmathlibs/Pages/default.aspx
In case of dense systems that's basically linpack:
http://code.compeng.uni-frankfurt.de/projects/hpl