Addendum was Re: now problems RUNNING molpro2002.6: ADDRESS ERROR crashes

Garold Murdachaew garold at physics.udel.edu
Fri Aug 1 19:57:19 BST 2003


Hi Matt and other molpro experts,

Thank you for your assistance.  A member of my group had recommended
using export LD_ASSUME_KERNEL=2.2.5 in the .bashrc file but I had 
commented it out in the expectation that the ifc / RH9 incompatibility 
fix recommended in 

	http://www.wien2k.at/reg_user/faq/redhat90.html

was sufficinet.  However, I did what you suggested: make veryclean, used 
your CONFIG file recommendations, and recompiled.  As before, compilation 
was trouble-free and the log showed no errors.  However, I again had 
"Adress Error" when I ran.  Actually, I do not know too much about ld or 
the LD_ASSUME_KERNEL variable (nor NPTL--what's that?) so I'll read up on 
this.

Let me know if you have other suggestions.  Is the only solution 
downgrading to Red Hat 8 (or earlier)?  Has no one managed to install 
molpro under Red Hat 9?

Thanks again,
Garold

ps: Hyper-threading is definitely diasbled (this is what 
/var/log/dmesg shows).


On 1 Aug 2003, The Matt wrote:

> Also, if you explore the NPTL issue, you might want make veryclean and
> remake the whole sucker with LD_ASSUME_KERNEL=2.4.1 in the environment. 
> Some programs (like molpro) may be more statically linked, so just
> trying to run molpro under LD_ASSUME_KERNEL might fail.
> 
> Of course, compiling under LD_ASSUME_KERNEL might fail too.
> 
> Of more course, you could have already done this.
> 
> Hope this can help,
> Matt
> 





More information about the Molpro-user mailing list