The ABINIT Users Mailing List ( CLOSED )

["Nuno A. G. Bandeira" <nuno.bandeira@ist.utl.pt>]

Paul Fons wrote:
> Yes this is the gnu 4.0 g95 fortran.  I gather it is a work in  
> progress, but it actually compiled the abinit source without error  and 
> passed the floating point tests.  I have no idea at this point  what 
> quality of the code generated by the g96 compiler (the absoft  compile 
> generated errors so speed is not the important point there).   In the 
> (included) comparisons with the P6 code in Tests_v1-v4, the  speed 
> varied from about the same as a P6 to about twice as fast (a  single CPU 
> of a 2.7 GHz G4).  

I have run the tests (tests_cpu, test_fast, Test_in, test_v1...) also 
with g95 compilates, albeit on a different platform, and the numerical 
efficiency and speed are impressive.

g95 is constantly being updated almost on a daily basis to remove bugs 
and upgrade efficiency and the people involved are also interested in 
using large codes to run it so it's a promising venture and it's now 
reaching maturity. Gfortran on the other hand is still in its infancy 
and can't even compile some f77 sources because they decided to rewrite 
from scratch the g77 code they inherited.


cheers,
--
Nuno A. G. Bandeira, AMRSC
Graduate researcher and molecular sculptor
Inorganic and Theoretical Chemistry Group,
Faculty of Science
University of Lisbon - C8 building, Campo Grande,
1749-016 Lisbon,Portugal
http://cqb.fc.ul.pt/intheochem/nuno.html
Doctoral student @ IST,Lisbon
--