[CPMD-list] Time for compute NMR chemical shifts

[Axel Kohlmeyer]

>>> "ST" == sttp komin <sttp_komin at hotmail.com> writes:

ST> Dear All,

ST> I run cpmd to calculate NMR in C6H6 from input file example.
ST> The lenght time which run on linux os with athlon XP 1.33 Mhz. and ram 256 
ST> MB free disk 2 Gbyte and IFC compiler  was  shown that
ST> CPU TIME :    0 HOURS 32 MINUTES  5.77 SECONDS
ST> ELAPSED TIME :    0 HOURS 50 MINUTES 11.97 SECONDS

ST> when  compare with the output file example shown that
ST> CPU TIME :    0 HOURS  1 MINUTES 48.95 SECONDS
ST> ELAPSED TIME :    0 HOURS  1 MINUTES 52.90 SECONDS

ST> So i would like to ask that  which computer which you use to test  the 
ST> output file?(parallel?) and how i can improve
ST> my  calculation time?

hi.

the example run _is_ a parallel run, as you can see for yourself
if you look for the string 'PARAPARAPARA' in the output file.
the following section shows how the job is distributed over the
parallel nodes. seems to be a 32-cpu parallel run, btw.

given the size of the box and the cutoff in the example the
job will have needed more memory than the ~200MB available
on your machine and i assume the machine has been swapping 
happily during the run. see the difference between cpu time
and elapsed time. of course i assume you have not run anything 
else on the machine at the same time.

so to improve (a bit) you need more memory. if you have not done
so already, you should use an optimized LAPACK/BLAS library.
this can be either build from the ATLAS package <http://math-atlas.sf.net>
or you can try the intel math kernel library (pentium-III) version.
an athlon-xp optimized binary of a combined LAPACK/BLAS/ATLAS library
can be found at: 
<http://www.theochem.ruhr-uni-bochum.de/~axel.kohlmeyer/libatlas.a>

to get into the neighborhood of 2 minutes you either need a large
multiprocessor smp-box (16 cpu or more) or a cluster of pcs with very
fast interconnect (dolphin/SCI or myrinet).


ST> and the last problem is  i can not run file input example to test full 
ST> calculation on NMR chemical shifts.
ST> The result is shown below.

ST> ****************************************************************
ST> *                    RESPONSE CALCULATIONS                     *
ST> ****************************************************************


ST> ********************************************************RESPONSE*
ST> * RESPONSE TYPE:           NMR chemical shifts                  *
ST> * PRECONDITIONER TYPE:     smooth (ds)                          *
ST> * PRECONDITIONER CUTOFF:   0.50000                              *
ST> * WFN CONVERGENCE:         0.100E-04                            *
ST> * GND STATE WFN:                      reciprocal space only     *
ST> * ALGORITHM:               preconditioned conj gradient         *
ST> *                          Fletcher-Reeves variant              *
ST> * CG: ANALYTIC STEPS:      3                                    *
ST> * CG: LENGTH PREFACTOR:    0.60000                              *
ST> ********************************************************RESPONSE*
ST> ***  DO_PERTU| THE NEW SIZE OF THE PROGRAM IS  170996 kBYTES ***

ST> ****************************************************************
ST> PROCESSOR     0 ALLOCATION OF    51260040 WORDS OF MEMORY FAILED
ST> ****************************************************************


ST> ***    MEMORY| THE NEW SIZE OF THE PROGRAM IS  171060 kBYTES ***

ST> ================================================================
ST>                       BIG MEMORY ALLOCATIONS
ST> scr             12974052                PME              4746380
ST> psi              2162342                YF               2162342
ST> PSI              2162342                XF               2162342
ST> CATOM            1898520                SCR              1587897
ST> GDE              1186595                rhoe             1081171
ST> ----------------------------------------------------------------
ST> [PEAK NUMBER   80]      PEAK MEMORY     29294612 =  234.4 MBytes
ST> ================================================================
ST> STOPGM! STACK OF MAIN CALLS:
ST> STOPGM! CALL        NMR


ST> PROGRAM STOPS IN SUBROUTINE MEMORY| ALLOCATION FAILED (C1)

ST> what should i do? or any one have any suggestion.

well, the error message says it: you need more memory!
the program tried to request 51260040 words (= ~400MB) and the
request failed.

regards,
        axel kohlmeyer.

ST> Thank you and with best wishes!
ST> Sincerely yours,

ST> Sittipong Komin
ST> Physics Dept.
ST> Science Fact.
ST> Ubonratchatani University
ST> 34190
ST> Thailand



--

=======================================================================
Dr. 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.