Hi, I can't remember the why, but I recall that with closed source software the incentives are stronger to have it static.
Any offers on this? It could have been that debugging conflicting APIs is totally avoided, since without source they are impossible to reconcile. F. Αποστολή από κινητό Samsung. <div>-------- Αρχικό μήνυμα --------</div><div>Από: Kenneth Hoste <[email protected]> </div><div>Ημερομηνία:2017/03/14 13:25 (GMT+02:00) </div><div>Προς: [email protected] </div><div>Θέμα: Re: [easybuild] Why is MKL linked statically and other BLAS not? </div><div> </div>Dear Bart, On 13/03/2017 13:43, Bart Oldeman wrote: Hi, in framework intelmkl.py it says: BLAS_LIB_STATIC = True (hardcoded, not an option) I'm not sure why this is only done for Intel MKL, and not for other BLAS/LAPACK libraries. This was done as a part of a *big* refactor years ago (when we still used the term 'toolkit' rather than 'toolchain'), see https://github.com/hpcugent/easybuild-framework/pull/281 . We can certainly enhance this so you get the choice, e.g. via a toolchain option named 'blas_link_static'? regards, Kenneth I did some "git blame"ing to find out the reason and found this commit introduced it: commit 65c5e3ea01f8a175244115629090ace41fa30fdc Author: stdweird <[email protected]> Date: Tue Sep 18 10:22:07 2012 +0200 blas added, need better Variables class (refs #176) --- this does not really answer my question. Does anybody on the list know. Note: all other linalg libraries (gotoblas, acml, openblas, atlas, libsci, flame, ...) are linked dynamically; MKL is the exception. Regards, -- Dr. Bart E. Oldeman | [email protected] | [email protected] Scientific Computing Analyst / Analyste en calcul scientifique McGill HPC Centre / Centre de Calcul Haute Performance de McGill | http://www.hpc.mcgill.ca Calcul Québec | http://www.calculquebec.ca Compute/Calcul Canada | http://www.computecanada.ca Tel/Tél: 514-396-8926 | Fax/Télécopieur: 514-396-8934
