As I said before, it is very likely one of the default values is not set, though I haven't checked this. This will have the effect of setting that value (what value I don't know) to 0, which is almost certainly invalid.
The solution is the same in all cases. Run the tuning program. Unless it is also buggy, you will get a valid set of values and won't have to rely on possibly erroneous or missing fallback values. I think you are confusing the value that you are setting from 11 to 0, both of which are valid, with *other* missing tuning values, which then default to 0. There may be a constraint that the *other* value needs to be greater than the value that was set to a default of 11. The tuning program knows this. Apparently the fallback values do not. Bill. On 30 September 2015 at 14:54, Bill Hart <goodwillh...@googlemail.com> wrote: > > > On 30 September 2015 at 14:29, Jeroen Demeyer <jdeme...@cage.ugent.be> > wrote: > >> On 2015-09-30 14:06, 'Bill Hart' via mpir-devel wrote: >> >>> for the two algorithms in >>> question there are no "invalid values". >>> >> >> Why do you say that? For the two variables in question, there are >> certainly invalid values: changing the default value from 11 to 0 fixes the >> problem. So I would say that the 11 is invalid (or maybe it's only invalid >> in combination with other values, but I don't think that matters much). > > > I don't see how this is relevant. > > >> >> The bug is in *all likelihood* caused by the tuning program never having >>> been run for those arches. >>> >> I don't think it is realistic to require for a MPIR release to run the >> tuning program on all arches. > > > And that is relevant why? > > >> >> That's the motivation for my question: I think the *real* bug is that >> there exist invalid values in the first place, not that the tuning program >> hasn't been run. > > > There's nothing I can do about that. > > Bill. > -- You received this message because you are subscribed to the Google Groups "mpir-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to mpir-devel+unsubscr...@googlegroups.com. To post to this group, send email to mpir-devel@googlegroups.com. Visit this group at http://groups.google.com/group/mpir-devel. For more options, visit https://groups.google.com/d/optout.