On 09 Nov 2007, at 19:16, Sebastien Maret wrote: > Maybe I am overlooking something, but why do you need ATLAS on 10.5? > Both BLAS and Lapack libraries (which I think is what ATLAS provides) > are present in /usr/lib since 10.4 (Intel). They were missing on 10.4 > PPC, so I filled out a radar about a year ago; Apple sent me an email > a few days ago saying that this was fixed in 10.5, but I haven't > checked. A couple of possible reasons : 1) Apple's libs have at some times been known to lack some symbols. 2) To have also the corresponding headers, if you want to compile against. 3) To have also the corresponding man pages, if you want to use the libs in your programs. 4) To have the choice between linking against an all-encompassing dylib or, if you program uses only a very small fraction of the code, against the .a archives, to gain some speed and diminish the footprint. 5) If it is to do scientific work, which has to remain check-able, to decrease the blackbox aspect. 6) To have libraries tuned to your own machine, with significant speed benefits expected. Was trying to time netlib's blas tests (in atlas' infotest part) against both, to illustrate the above, against both dylib's, but with /usr/lib/libblas.dylib, but while xblat1s and xblat1d had no problems, it ran into a seg-fault with the complex counterparts ... Avoiding that type of trouble may be an additional reason ....
Jean-Francois ------------------------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/ _______________________________________________ Fink-users mailing list Fink-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fink-users
