On Fri, Dec 30, 2011 at 3:25 PM, Alan W. Irwin <ir...@beluga.phys.uvic.ca>wrote:
> I have investigated further, and I now think Dave is right with his > desire to go ahead with the release of the rc2 versions of the > lapack/blas find modules. (Sorry, Dave, for this flip-flop at the > last minute, but such deeper investigations take time.) > > What appears to be going on is that in 2.8.5 the usual mechanisms to > choose libraries (CMAKE_LIBRARY_PATH, CMAKE_FIND_ROOT_PATH) work > correctly because they trump (according to my tests) the vendor choice > (which defaults to ALL). > So even though my system has both an atlas and generic version of blas > installed, that > is ignored for 2.8.5 and the generic blas version is used instead from > the location designated by CMAKE_LIBRARY_PATH or CMAKE_FIND_ROOT_PATH). > > But in 2.8.7-rc2, the vendor choice trumps CMAKE_LIBRARY_PATH and > CMAKE_FIND_ROOT_PATH. So given a choice between the generic version > of blas in CMAKE_LIBRARY_PATH and my system's generic and atlas > versions of blas, rc2 picks the system atlas version. I guess that is > a reasonable choice because for large matrices atlas is normally > faster than the generic version. However, for small matrices (such as > in FreeEOS) it is noticeably slower. > > So in sum, there has been a fundamental shift in how the lapack/blas > libraries are chosen. I assume that was a deliberate change in design > so it was probably unfair for me to call it a regression, but a > prominent announcement of this important change would have been > appropriate to avoid misunderstandings. > > For my own use case, I have found setting -DBLA_VENDOR:STRING=Generic > and CMAKE_LIBRARY_PATH (or presumably CMAKE_FIND_ROOT_PATH which > I haven't tested for this configuration) gives me > > > -- LAPACK_LIBRARIES = > /home/software/lapack/install_**double/lib/liblapack.a;/home/** > software/lapack/install_**double/lib/libblas.a > > i.e., the results I want for FreeEOS both 2.8.5 and 2.8.7-rc2. > > That'a slightly more complicated for FreeEOS users (which tend to have > their own built versions of lapack/blas just like I do) than what > worked up to and including CMake-2.8.5. But that additional > complication is completely acceptable. > > > Alan > __________________________ > Alan W. Irwin > > Astronomical research affiliation with Department of Physics and Astronomy, > University of Victoria (astrowww.phys.uvic.ca). > > Programming affiliations with the FreeEOS equation-of-state > implementation for stellar interiors (freeeos.sf.net); the Time > Ephemerides project (timeephem.sf.net); PLplot scientific plotting > software package (plplot.sf.net); the libLASi project > (unifont.org/lasi); the Loads of Linux Links project (loll.sf.net); > and the Linux Brochure Project (lbproject.sf.net). > __________________________ > > Linux-powered Science > __________________________ > I don't know about the rest of you here, but I certainly welcome flip-floppers when they flip over to my side of a debate. :-) 2.8.7 is imminent!
-- Powered by www.kitware.com Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html Please keep messages on-topic and check the CMake FAQ at: http://www.cmake.org/Wiki/CMake_FAQ Follow this link to subscribe/unsubscribe: http://www.cmake.org/mailman/listinfo/cmake
