Re: gfortran

[Dominique Rousset <dominique.rousset@free.fr>]

Le Lundi 24 Octobre 2005 23:27, Dominique Rousset a écrit :
> Le Lundi 24 Octobre 2005 23:00, Lennart Sorensen a écrit :
> Further investigations indicate an extra time step on AMD64 which result in
> one extra floating point value per record.
> The number of time steps is computed using floating point arithmetics, so
> very small differences may result in a difference of 1.
> The results may be correct.
> I cannot check now.

Ok, it's really a a matter of accuracy and truncation. (and not a question of 
I/O)
From the algorithm point of view, debian amd64 seems to be the accurate one. 

I just find a little strange that the difference did not appear on SUN 
UltraSPARC and IBM PowerPC which are also 64 bits processors.


D.


-- 
Dominique Rousset
dominique.rousset@free.fr