I suppose it does not help that Apple is moving over to Metal from OpenGL. Do the offending modules have to do with OpenGL?
On Tue, Jun 9, 2015 at 7:00 PM, Alexander Hansen < alexanderk.han...@gmail.com> wrote: > > > On Jun 9, 2015, at 18:48, Sunil Shah <ssha...@gmail.com> wrote: > > > > I should have been careful in my statement about what worked for me. > > > > I am trying to understand the problem you referred to about octave 4.0.0 > not building on clang++. > > > > I have been quite impressed with the performance of Octave code on Mac > OSX using fink. My tests used octave382-atlas-qtmac 3.8.2-6. It has also > been quite reliable. > > > > The fact that the code works much faster on four year old hardware with > a lot fewer cpu cores and even after using the latest Ubuntu tools / > libraries on the high end Intel machines on EC2 is remarkable. > > > > Apple does claim that its Accelerate framework is much faster than > standard BLAS / ATLAS. A lot of the improvements were part of 10.9 / 10.10 > update (see WWDC 2013-2014 videos). I will check if the default "octave" > performs faster than the octave-atlas version. > > > > Now, as you imply in your last email, if the default "octave" variant > uses clang already, what is the issue with octave 4.0.0 and clang++? Or is > it a new issue in octave 4.0.0 in moving from 3.8.2? > > > > > > Almost everything in Fink uses clang/clang++ to build on 10.7 and later. > I don’t happen to have a transcript handy, but it’s a new issue in moving > from 3.8.2, and the issue is that Apple’s clang doesn’t like some of the > source code and thus Octave can’t be built. > -- > Alexander Hansen, Ph.D. > Fink User Liaison > >
------------------------------------------------------------------------------
_______________________________________________ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
