On 2012-06-06 09:42, Georg S. Weber wrote: > the only thing I could think of is the problem with building Sage > binaries (I've got myself a Quad Core i7, i.e. "Sandybridge" > architecture) and wanting to use these on other people's Macs (which > might have "only" a Core2Duo CPU). Till Sage 4.8, this wasn't a probem, > because the compiler(s) used didn't support Nehalem/Westmere or > Sandybridge/Ivybridge optimizations. Since Sage 5.0, we use gcc 4.6 > which does have support --- but no infrasctructure in place to bail out > if a Sage binary, that is built on a Sandybridge CPU with the respective > optimizations, is installed a a Mac with a CPU older than that. > (On Linux, some "sage_flags.txt" fie is used for that, but might have > problems with Sandybridge, too --- bringing possibly optimizations for > SSE4.1/SSE4.2 (POPCNT) and AVX extensions, which are not tested for > currently). What you can do is set the "SAGE_FAT_BINARY" environment > variable, that should work somehow also on Macs in the oncoming Sage > v5.0.1 version (but possibly leads to pretty slow binaries, compared to > optimized ones).
In short: if you want to move your binary to a different computer, you should compile Sage with SAGE_FAT_BINARY=yes and it should "just work" in sage-5.0.1.rc0. Except for Linux x86, such a binary will run slower than a natively compiled one. -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
