On Jan 13, 2:38 am, ja...@njkfrudils.plus.com wrote: > > > The other issue here is that there are multiple algorithms here and it > > > is quite likely that very small examples will use the basic xgcd which > > > may well guarantee minimality. Only if you have say hundreds of > > > thousands of bits will the algorithm used not return minimally > > > guaranteed results. I'm sure when the time comes we can give some > > > explicit million digit integers for which minimality is not > > > guaranteed. > > The non-minimal examples could be dependant on tuning parameters that select > between different algorithms, making this sort testing impratical.
That's painful. One of the great selling points for MPFR is that it always gives the exact same answers on all platforms; it would be sad if MPIR doesn't provide the same guarantees. How much speed difference are we talking about? IMHO, I would accept a few percent slowdown for huge GCDs on some platforms to get portable, identical results cross-platform in Sage. (Of course, it's not my decision to make.) Carl --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "mpir-devel" group. To post to this group, send email to mpir-devel@googlegroups.com To unsubscribe from this group, send email to mpir-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/mpir-devel?hl=en -~----------~----~----~----~------~----~------~--~---
