[CPMD-list] help on compiling

[Axel Kohlmeyer]

>>> "HW" == HW Sheng <hwsheng at jhu.edu> writes:


HW> Hi, Axel.

hi howard,

HW> As instructed, I checked the data segment and stack size limit using
HW> "ulimit -a". It reads like this:

HW> core file size (blocks)        0
HW> data seg size (kbytes)        unlimited
HW> file size (blocks)        unlimited
HW> max locked memory (kbytes)        unlimited
HW> max memory size (kbytes)        unlimited
HW> open files                  1024
HW> pipe size (512 bytes)       8
HW> stack size (kbytes)         8192
HW> cpu time (seconds)          unlimited
HW> max user processes          2046
HW> virtual memory (kbytes)     unlimited

HW> The stack size is much smaller than yours. But I am not sure whether or not
HW> this will have a catastrophic effect on running cpmd jobs. Your thoughts?

it may cause crashing jobs in some cases, because the compaq fortran
compilers use the stack extensively. but since cpmd uses dynamical
memory management, i suspect it will happen more often to older fortran
code with static arrays. anyway, i recommend increasing the
stacksize just in case. if you stack is too small, your jobs will
usually crash with a segmentation fault (SIGSEGV) or a bus error (SIGBUS).

BTW, if you want to have it changed permanently for all users
you should, on linux machines, modify the file /etc/security/limits.conf
accordingly.

HW> By the way, are you also using compaq fortran 1.2 to compile cpmd? did you
HW> see the similar problem (see below) while compiling?

we have 1.1.1. last time i looked, 1.2 was still beta, but i have not
looked in a while.

HW> f90: Info: ./swap.inc, line 28: Symbol in BLOCK DATA program unit is not in
HW> a COMMON block. [IUNITSW]
HW>     INTEGER       ISWFIRST, NSWBLOCK,IUNITSW,ISIDISW,ISWRES
HW> ....

HW> (basically, it's indicating that the definition of line 28 and line 51 in
HW> swap.inc is not in a common block. Here, I use "fort" instead of mpif77. In
HW> fact, if mpif77 is invoked, this problem persists).

AFAICT, this is just an informational message. compiling cpmd can cause
many of those on certain platforms, some give hints about problems in
the sourcecode, but some are warnings about perfectly correct code that may
experience a performance degradation or is considered bad programming
style by the compiler authors.

HW> Thank you.

HW> Howard

you're welcome.  cheers,
       axel.

HW> PS: your ideas about TCPU I posted yesterday are very plausible. I'll test
HW> the timing results following your suggestions.



HW> _______________________________________________
HW> CPMD-list mailing list
HW> CPMD-list at cpmd.org
HW> http://www.cpmd.org/mailman/listinfo/cpmd-list



--

=======================================================================
Axel Kohlmeyer       e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de
Lehrstuhl fuer Theoretische Chemie          Phone: ++49 (0)234/32-26673
Ruhr-Universitaet Bochum - NC 03/53         Fax:   ++49 (0)234/32-14045
D-44780 Bochum                   http://www.theochem.ruhr-uni-bochum.de
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.