But the command is mostly linear algebra over QQ. Its possible that the old GCC does something very stupid, but I doubt that you can speed up integer computations by a factor of 5x just by optimizing code paths. So it pretty much must be the blas inside linbox.
On Monday, January 7, 2013 9:18:23 PM UTC, Jeroen Demeyer wrote: > > On 2013-01-07 21:59, Volker Braun wrote: > > Its pretty clear that the newer GCC correctly vectorizes ATLAS for the > > CPU whereas the old one did not. > > I doubt it's ATLAS. I didn't find any particular difference in the ATLAS > install logs. It's curious to see that ATLAS somehow brute-forces CPU > flags, it even tries stuff like -mcpu=ultrasparc on x86! > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To post to this group, send email to sage-devel@googlegroups.com. To unsubscribe from this group, send email to sage-devel+unsubscr...@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en.
