Okay, seems like the build I had of g95 was duff, I got a new build today
(28th March) and the following are what I see:
1) Landau (32 bit P4)
g95 x86 32 bit no MKL = Pass
g95 x86 32 bit MKL 8.0.2 = Pass
2) Caffeine (32/64 bit Dual Pentium D)
g95 x86_64 - 32 bit no MKL = Pass
g95 x86_64 - 32 bit MKL 8.0.2 = Pass
g95 x86_64 - 64 bit no MKL = Fails on cytosine and all LES_GB
tests
- testing ended at ptraj with a seg fault. See attached Log file.
g95 x86_64 - 64 bit MKL 8.0.2 =
So it would seem that g95 is fine as long as one only tries to build 32
bit
executables either on an x86 machine or an x86_64 machine. If you try and
build 64 bit executables most things will work (if you don't use
MKL)except
for Cytosine (don't know why), all the LES_GB test cases and ptraj
(although
isn't ptraj all in c?). Everything else seems to work. If you use MKL then
targetted MD fails with a DGEEV error and all of the divcon tests also
fail
with Double Linked List Corruption Errors???
I don't know whether we should worry about this or not. Perhaps we should
take a look through the LES GB code or perhaps we should just assume it is
a
compiler bug that may get fixed at some point.
All the best
Ross
/\
\/
|\oss Walker
| HPC Consultant and Staff Scientist |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
|
http://www.rosswalker.co.uk | PGP Key available on request |
Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.
Received on Wed Apr 05 2006 - 23:49:33 PDT
