[MAP] MARS15(2012) on BNL computers

[Yip, Kin]

Hi Nikolai,

rlnxsp01/2 has AFS.   Eg. if I do the following :

setenv ROOTSYS                 /afs/rhic.bnl.gov/star/ROOT/5.34.03/.sl53_x8664_gcc432/rootdeb
setenv LD_LIBRARY_PATH /afs/rhic.bnl.gov/star/ROOT/5.34.03/.sl53_x8664_gcc432/rootdeb/lib:$LD_LIBRARY_PATH

root.exe would work.

Do you think you may use this in rlnxsp machines ?    For ROOT in cluster/spot, we have to install or ask to install ...

Kin

> -----Original Message-----
> From: Nikolai Mokhov [mailto:mokhov at fnal.gov]
> Sent: Wednesday, December 26, 2012 10:13 PM
> To: Harold Kirk; Ding, Xiaoping; Yip, Kin
> Cc: MAP-l at lists.bnl.gov
> Subject: MARS15(2012) on BNL computers
> 
> Hi all,
> 
> I have successfully installed and tested the newest MARS15(2012) of December 2012 on
> BNL's
> cluster/spot in /home3/nmokhov/restricted/mars15
> and rlnxsp01 in /u0b/mokhov/restricted/mars15
> 
> The only new feature I could not check on both the clusters is the ROOT geometry,
> because I couldn't find the required ROOT version 5.32 or 5.34 installed there. Let me
> know once it is there and I will add this powerful feature.
> 
> I have checked all other new modules including EGS5 and ENDF/B.
> 
> 1. Update your .bashrc in accordance with that in my home directory (MARS, EGS and
> ENDF/B parts; we would need to define there corresponding ROOT variables once the
> ROOT is in place).
> 
> 2 . Copy to your area GNUmakefile, *INP*, m1512.f, marsmain.f and xsdir files as well
> as the auxiliary directory. The latter contains the newest manual, tutorials, other
> documentation along with three "How-to" subdirectories on: nuclide inventory; preparation,
> submission and averaging results of multiple jobs; and rather generic .bashrc example.
> 
> As always, to reproduce a sample problem of that in my directory just say
> 
> make
> time ./rmars-bnab-linux &            for a basic mode
> 
> or
> 
> make rmars-mcnp-linux
> 
> time ./rmars-mcnp-linux &            for a mode with ENDF/B  x-sections for neutrons
> below 14 MeV
> 
> To activate the ROOT or/and EGS5 modules, uncomment corresponding lines at the
> beginning of GNUmakefile.
> 
> Good luck!
> 
> Nikolai
> 
> 
> -----Original Message-----
> From: Harold Kirk [mailto:hkirk at bnl.gov]
> Sent: Friday, December 21, 2012 9:16 AM
> To: Nikolai Mokhov
> Subject: FW: Checking mars15
> 
> Nikolai,   We have had a significant upgrade of the BNL ITD cluster
> software.  As you can see below, this has resulted in a complete cessation of operations for
> the MARS efforts on this machine.  It appears that the best solution is for you to rebuild.
> Please look into this a soon a practical as we are currently dead in the water with our
> MARS simulations work.  Thanks.
> 
> 
> Harold G. Kirk                               e-mail: hkirk at bnl.gov
> Dept. of Physics, 901A                Voice:  (631) 344-3780
> Brookhaven National Lab           Fax:    (631) 344-3248
> Upton, NY 11973-5000                http://pubweb.bnl.gov/people/kirk
> 
> -----Original Message-----
> From: Ding, Xiaoping [mailto:xding at bnl.gov]
> Sent: Friday, December 21, 2012 9:47 AM
> To: Berg, J Scott; Harold Kirk; Kamal Sayed Abdalgaffar, Hisham
> Subject: RE: Checking mars15
> 
> Hi Harold,
> 
> Could you please send an request to Mokhov to rebuild his mars15 on BNL new system?
> 
> Thanks,
> 
> Xiaoping
> 
> ________________________________________
> From: Berg, J Scott
> Sent: Wednesday, December 19, 2012 9:28 AM
> To: Harold Kirk; Ding, Xiaoping; Kamal Sayed Abdalgaffar, Hisham
> Subject: Checking mars15
> 
> First of all, mars15 was built with the gcc/gfortran 4.1 compilers, whereas the new system
> has the gcc/gfortran 4.4 compilers. There is a fortran library compatability issue which
> prevents a successful link of the new library. One could have ITD install the gfortran 4.1
> compatability library, but this has a good chance of creating problems due to linking
> against two fortran libraries which may or may not get along well. There's also another
> problem with missing libraries in Nikolai's tk version, but that is fixable by either installing
> the missing library or just using the system tk library. I think Mokhov really needs to
> rebuild on the new system.
> 
> One temporary workaround is to install a Scientific Linux 5 VM on your desktop/laptop,
> and build your executables there.
> 
> Three of the mcnp datasets seem to be inaccessible: uresa and xsdir.
> Intentionally or not, read access for "others" has been removed. These sets are not in
> Mokhov's directory; he just has a link to the actual location, which is a user kinyip.
> 
> -Scott
>