On Thu, Aug 4, 2016 at 11:17 AM, leif <not.rea...@online.de> wrote: > Jean-Pierre Flori wrote: >> On Thursday, August 4, 2016 at 11:07:38 AM UTC+2, Erik Bray wrote: >> >> Tangentially related, but it might be nice if GMP/MPIR got the same >> treatment that some other packages have been getting, a few at a time, >> of using the system version if it's available and current enough. >> Really should be done for every package in fact, but that's something >> I'm happy to take one at a time, especially for such core packages.... >> >> Note that GMP uses a lot of assembly when properly built and is used >> basically by all other libraries. >> Using a prebuilt one without specific assembly is maybe not the best idea. > > The assembly is not a problem, since most distros build GMP with > --enable-fat. But the rest gets compiled with arch and tune generic, > not exploiting any instruction set extensions most CPUs nowadays have. > > But we could at least /offer/ the option to build Sage with > --with-[g]mp=system etc.
That's not even necessarily all true. Just because it's the "system" GMP doesn't mean it hasn't been tuned. By default, no, but most OS's have instructions for building tuned system installs of MP, BLAS, etc. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at https://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
