Re: .f90 runs slower than .f

[Rachel Dowdall <Rachel dot Dowdall at glasgow dot ac dot uk>]

Thank you for the replies. I have tried to provide all the additional 
info below. I apologise if this turns out to be due to my stupidity!

> It would help enormously if you would provide a driver main program so
> that we can test your assertion.
I have put a tar file of my code in my dropbox:
https://dl.dropbox.com/u/70250131/gauge_field_v0.44-gfortrancheck.tar.bz2

I have tried to get rid of as much junk as possible but I decimated it 
slightly trying to work out what was going wrong. Unfortunately the 
smallest data file I have is 50Mb which I have had to include in the 
directory.
There are two options in the makefile: NRQCDf90 and NRQCDf77 differing 
only by whether the file psq.f or psq.f90 is included. These are 
identical except for the file extension. Once made, then run with
./NRQCDf90 simple.xml
./NRQCDf77 simple.xml
The speeds are visibly different.

>   If I have understood correctly, the
> time that you are quoting includes compilation?  Or is it just the run
> time?
Sorry I wasn't clear, the time was just the output from
time ./NRCQDf90.xml

I am using:
GNU Fortran (GCC) 4.7.0 20110528 (experimental)
on Ubuntu 10.04. My cpu info is in an attached file.

I have also attached the output of the profiler in both cases although I 
cannot see how this helps. I have turned off most of the other functions 
so that the run time is dominated by the psq routine (which is pretty 
much true anyway.) In case it is not clear from the psq code, it is 
doing a second derivative on a lattice of 3x3 complex double precision 
matrices.

Thanks
Rachel

--
===================================
Dr Rachel Dowdall
School of Physics and Astronomy
University of Glasgow
Glasgow G12 8QQ
email: Rachel.Dowdall@glasgow.ac.uk
tel: +44 141 330 3770
===================================

Attachment: profilef77.txt
Description: Text document

Attachment: profilef90.txt
Description: Text document

Attachment: cpuinfo
Description: Text document