Hi, What's the situation with speed regression testing in Sage? A couple of years ago I think people wrote a regression testing framework (maybe David Roe or Robert Bradshaw?)
MOTIVATION: I ask because at Sage Days 61 I keep hearing about how our exact linear algebra in Sage is terrible (with remarks like: 'maybe we need to look into writing our own fast linear algebra'... a road actually well traveled in Sage many years ago), how people are having to use Magma for basic modular symbols computations instead of Sage due to linear algebra speed, and there is this massive patch http://trac.sagemath.org/ticket/16803 about fixing the ridiculously slow dense matrices over ZZ -- which used to be ridiculously fast ?!. So I started looking into 16803 and looks to me like the function _multiply_linbox is like 10-20 times slower than it was when we introduced it as the default for matrix multiplication over ZZ. (see the ticket for timings). I suspect that we're linking linbox incorrectly, which could easily result in this sort of slowdown, since linbox uses a floating point BLAS at a certain point. If we mess up linbox, then tons of stuff all over sage involving linear algebra would be slowed down as a result. E.g., we use linbox for charpoly computations. Anyway, there are a ton of people working on Sage, and even very tiny changes in one part can have speed impacts in far away places... so it would be good if we were to run a speed regression test on each version, compare timings, see if anything dropped in a new version, etc. Does anybody have any relevant code...? That said, regarding 16803, switching to building our dense matrix algebra over ZZ on one single clean modern C library is a great improvement (thanks FLINT'ers and Marc), and if some regressions motivated that, I'm not complaining! William -- William Stein Professor of Mathematics University of Washington http://wstein.org wst...@uw.edu -- 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 http://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
