RIP4 on a Xeon EM64T - Repost from old forum

The NCL graphics package.

RIP4 on a Xeon EM64T - Repost from old forum

Postby malonso » Sat Nov 29, 2008 11:47 pm

By changing the Makefile a little I've managed to compile RIP4 with version 6.0 of the Portland Group compiler on a dual XeonEM64T running Fedora Core 2.

Code: Select all
LIBS    = -L$(NCARG_ROOT)/lib -lncarg -lcgm -lncarg_gks -lncarg_c [b]-L/usr/X11R6/lib64[/b] -lX11 -L$(PGI)/linux86/lib -lpgftnrtl -lpgc [b]-L/usr/lib64 -lg2c[/b]  -L/usr/local/unidata/lib -lnetcdf


I can use ripdp_mm5 successfully but when I run rip I get a segmentation fault almost immediately. (e.g.)

Welcome to your friendly RIP output file :
Temp elevation correction flag:
Lapse rate for correction: 6.4960001E-03
Color Names and Fractions Defined
Get header info.
Model info: V3.7.0 Grell MRF PBL Reisner 1 20 km, 25 levels, 60 sec

Requested time 0.000000 is available.
Segmentation fault (core dumped)

Has anyone successfully run rip on a similar setup or could suggest a solution.

Cheers
Graham
malonso
Site Admin
 
Posts: 158
Joined: Sun Mar 02, 2008 9:48 pm

Re: RIP4 on a Xeon EM64T - Repost from old forum

Postby malonso » Sat Nov 29, 2008 11:47 pm

Not sure about your particular case but I have found cases where rip failed, and doing an "unlimit" before running it fixed the problem. It is presumably related to memory allocation of some of the arrays in subroutines (my guess).

Brian
malonso
Site Admin
 
Posts: 158
Joined: Sun Mar 02, 2008 9:48 pm

Re: RIP4 on a Xeon EM64T - Repost from old forum

Postby malonso » Sat Nov 29, 2008 11:50 pm

Graham:

I have the same problem on my dual 64-bit Opteron running SuSE 9.1.

Code: Select all
$ rip WRF_output rip.in
 Welcome to your friendly RIP output file :
 Color Names and Fractions Defined
 Get header info.
 Model info: V3.6.0  Grell     MRF PBL    Simple ice   36 km,  32 levels,  108 sec

 Requested time     48.00000      is available.
Segmentation fault (core dumped)



Problem is the same using WRF or MM5 output. Ripdp_mm5 and ripdp_wrf work nicely, but rip fails with this seg fault. I increased my limits, as Brian recommends, but with no success.

I know it hasn't been long since your post, but did you ever find a solution? If so, I'd love to know what it was.

Thanks for any hints, suggestions, etc.
-matt
malonso
Site Admin
 
Posts: 158
Joined: Sun Mar 02, 2008 9:48 pm

Re: RIP4 on a Xeon EM64T - Repost from old forum

Postby kwthomas » Sat Nov 29, 2008 11:50 pm

Has anyone that has had the problem tried compiling a 32-bit version, assuming that the OS supports it? Compiling a program that works in 32-bit mode as a 64-bit program isn't guaranteed to work. Probably points of failures are integers that *MUST* be 32-bits, however, due to the way they are declared, they become 64-bits. I'm sure there are other reasons.

Kevin Thomas
Center for Analysis and Prediction of Storms
University of Oklahoma
Norman, Oklahoma
Kevin W. Thomas
Center for Analysis and Prediction of Storms
University of Oklahoma
kwthomas
 
Posts: 219
Joined: Thu Aug 07, 2008 6:53 pm


Return to NCL

Who is online

Users browsing this forum: No registered users and 1 guest