Changing the rewrite order greatly improves performance (though not yet to an acceptable level). I tested this using Singular's web-interface. For the inhomogeneous cyclic-8,
> int RT = rtimer; int T=timer; size(sba(k,0,0)); rtimer-RT; timer-T; 372 6369 5304 For the homogeneous cyclic-8, > int RT = rtimer; int T=timer; size(sba(k,0,0)); rtimer-RT; timer-T; 1182 6854 5113 There's another option for the module order, but I haven't found a setting that makes that improve. I find this curious, as I had the impression from conversations that things were much better than this. In fact, I thought they chose the default for the second option because experimental evidence suggested it was better, but here we see it getting worse. (I can certainly understand why the second option would be worse in practice, but then it shouldn't have ended up as the default. Then again, one example does not a standard make.) I have inquired & will report back. john perry On Sunday, May 8, 2016 at 12:56:32 AM UTC-5, parisse wrote: > > > > Le dimanche 8 mai 2016 04:08:54 UTC+2, john_perry_usm a écrit : >> >> What about homogeneous cyclic-8? I'm not sure it will be any better; I'm >> just curious. >> >> I do know Singular is working on improving aspects of the sba() >> implementation, and I'm a bit surprised it's that slow. >> > > That's indeed better : about 1mn15s (and 635M). But it's slower than > groebner (26s, 4M) and much slower than giac (3.6s). > -- 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.
