I just checked the FLINT implementation and there is this note: /* TODO: optimise this */ static void nmod_mat_addmul_classical(nmod_mat_t C, const nmod_mat_t A, const nmod_mat_t B)
So my guess is at the moment we wouldn't be competitive everywhere. I think the times we really win are when the modulus is "big". I have a vague recollection of a discussion about this issue where I noted that even for p = 50000, dim = 100 Sage suffered a > 100 times slowdown due to switch to very slow generic code. At p = 40000, dim = 100, my reading of the code was that a basecase implementation in Cython was used in Sage, not linbox (I couldn't see that it was ever used for matrix multiplication). I don't know how it compares with linbox, or why someone wrote a special Cython implementation. However, I also recall that in the discussion I have in mind, my abilities at reading code and understanding what Sage actually does were very questionable (and questioned). So I might be completely off with all of this! Take it with a grain of salt. Bill. On Jan 17, 11:13 am, Bill Hart <goodwillh...@googlemail.com> wrote: > At the moment we don't use the BLAS, so linbox is still better. But I > think Fredrik had some figures which show we beat Sage. > > Moreover, our code works for primes up to 64 bits which isn't possible > with the BLAS anyhow. > > I'll let Fredrik respond in more detail as he wrote the code. > > Bill. > > On Jan 17, 10:37 am, Martin Albrecht <martinralbre...@googlemail.com> > wrote: > > > > > > > > > Congratulations. I'm curious: how does your mod p multiplication of matrices > > compare ot LinBox's? > > > Cheers, > > Martin > > > -- > > name: Martin Albrecht > > _pgp:http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x8EF0DC99 > > _otr: 47F43D1A 5D68C36F 468BAEBA 640E8856 D7951CCF > > _www:http://martinralbrecht.wordpress.com/ > > _jab: martinralbre...@jabber.ccc.de -- 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
