Charles Greathouse on Tue, 10 Dec 2013 22:38:27 +0100


[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: Failed bench linear


Sorry. This was the latest git version as of when I sent my email:
http://pari.math.u-bordeaux.fr/cgi-bin/gitweb.cgi?p=pari.git;a=commitdiff;h=22a500468def5be6558baf8ae260dd64e6bf72bb


This was just make bench, no -j.

Charles Greathouse
Analyst/Programmer
Case Western Reserve University


On Tue, Dec 10, 2013 at 4:24 PM, Bill Allombert <Bill.Allombert@math.u-bordeaux1.fr> wrote:
On Tue, Dec 10, 2013 at 12:47:41AM -0500, Charles Greathouse wrote:
> My compiled gp-sta comes up with a strange answer for the lindep bench:
>
> ? lindep(Mod(1,7)*[2,-1;1,3])
> %1 = [Mod(6, 7), Mod(5, 7)]~
>
> which does not match the expected [-3, 1]~ from src/test/64/linear or my
> dynamically-compiled version of gp. Any idea what might be wrong? Any other
> information I should provide? The whole contents of the dif file:

Always report what says \v

> *** ../src/test/64/linear 2013-12-10 00:28:06.000000000 -0500
> --- gp.out 2013-12-10 00:35:18.000000000 -0500
> ***************
> *** 12,18 ****
>   ? charpoly(Mod(1,8191)*[1,2;3,4],z,2)
>   z^2 + Mod(8186, 8191)*z + Mod(8189, 8191)
>   ? lindep(Mod(1,7)*[2,-1;1,3])
> ! [-3, 1]~
> --- 12,18 ----
> ! [Mod(6, 7), Mod(5, 7)]~
>
> I think this is my first experience with a seriously wrong bench, rather
> than just precision errors or the like.

It is not "seriously wrong": if you look carefully, you will see that
both answer are correct.

This looks like a race condition. Did you do 'make bench -j' ?

Cheers,
Bill.