[Pw_forum] Example21 Problem

Axel Kohlmeyer akohlmey at cmm.chem.upenn.edu
Mon Jul 2 05:37:49 CEST 2007


On Sun, 1 Jul 2007, Amos Leffler wrote:

AL> Dear Forum,
AL>        I am unable to compile the program using the commands, 
AL> "-ftrace=frame" or "-ftrace=full" or I dont know how to formulate the 
AL> command properly.  The response is "no such command".

amos,

these are not commands but flags to the g95 compiler.
these kind of problems are not specific to QE. 
to get over this, i recommend you get some sort of
linux/unix training so that these rather elementary
issue don't get in the way in the future (there are 
plenty of books, online resources, workshops, linux
user groups, knowledgable colleages/friends etc.).

AL>        I have approximately 28 Gb empty space on my disk and 256 Mb of 
AL> RAM.  I always assumed that if more than 256 Mb is needed there will be 
AL> paging to the swap space.  This would slow the program but not stop it.  

you need to have about 1.5GB of swap space configured. this has to be 
done explicitely. use the 'free' command to verify it. here is the 
output from my laptop (512MB + 1GB swap):

[akohlmey at zero akohlmey]$ free -t
             total       used       free     shared    buffers     
cached
Mem:        510888     494480      16408          0      15996     
302508
-/+ buffers/cache:     175976     334912
Swap:      1020116      61892     958224
Total:     1531004     556372     974632

that being said, any computational code will perform horribly,
if you have to go into swap. unlike 20 or so years ago, disk
space is now relatively much much slower compared to RAM, so
using swap is not much of an option. with 256MB RAM you'll have
at best about 200MB left to run cp.x in (depending on what else
you have running on your machine and whether you are running 
X or not). that will bring your machine to a grinding halt
already with the 32 water example and you will run out of
address space with the 64 water (i assume you are running with
a 32bit OS).

AL> I checked "ulimit" and it replied unlimited.
AL>        At this point I have no idea of what is needed.  Hopefully 
AL> someone can help.

fist of all, you need more physical memory or else you'll
kill you machine (it is are desktop and not a notebook i
hope?). better yet you run on a parallel  machine (cluster).

cheers,
   axel.

AL>                                                                          
AL>          Amos Leffler
AL> 
AL> Axel Kohlmeyer wrote:
AL> > On Fri, 29 Jun 2007, Amos Leffler wrote:
AL> >
AL> >
AL> > amos,
AL> >
AL> > how much memory do you have on the machine that you are running on?
AL> > how large are your ulimit settings?
AL> >
AL> > the 32 water example requires about 1.2GB address space and 
AL> > about 400MB physical memory to run. for the larger systems
AL> > accordingly more. 
AL> >
AL> > when a job stops complaining about not being able to allocate
AL> > memory, it is likely that you either do not have enough memory
AL> > or address space available, or that your account settings 
AL> > prohibit that.
AL> >
AL> > cheers,
AL> >    axel.
AL> >
AL> > AL> Dear Forum,
AL> > AL>        Attached is the result of the following input in example 21:
AL> > AL>                    ./run_example 256
AL> > AL> It calculates values for all sizes up to 256 and skips the remainder.  
AL> > AL> However, I am puzzled that the calculations take very little time 
AL> > AL> certainly less than a minute for each one.  The "README" note says it 
AL> > AL> takes a long time and should include more than what is shown.  There may 
AL> > AL> be more to this than I understand.
AL> > AL>                                                                       
AL> > AL> Amos Leffler
AL> > AL> Axel Kohlmeyer wrote:
AL> > AL> > On 6/27/07, Amos Leffler <amos at errno.com> wrote:
AL> > AL> >> Dear Forum,
AL> > AL> >>        I found later after sending this note that the program ran (in
AL> > AL> >> background) but did the calculation for all of the numbers of water
AL> > AL> >> molecules.  The original input did have as the last entry a number (64)
AL> > AL> >> so there still is a problem.
AL> > AL> >
AL> > AL> > i disagree. please explain first the existence of the -ftrace=full
AL> > AL> > which is obviously causing the problems you see. omit it and
AL> > AL> > the script will run fine.
AL> > AL> >
AL> > AL> > axel.
AL> > AL> >
AL> > AL> >>
AL> > AL> >> Amos Leffler
AL> > AL> >>
AL> > AL> >> Axel Kohlmeyer wrote:
AL> > AL> >> > On Tue, 26 Jun 2007, Amos Leffler wrote:
AL> > AL> >> >
AL> > AL> >> > AL> Dear Forum,
AL> > AL> >> > AL>          Attached  are the details of a problem I ran into 
AL> > AL> >> trying to
AL> > AL> >> > AL> execute  example21.  I am using SuSE  Linux 10.2 which has 
AL> > AL> >> worked well
AL> > AL> >> > AL> up to this point so I think the problem is in the  run_example 
AL> > AL> >> file.
AL> > AL> >> >
AL> > AL> >> > well, i would say the problem is in your using of -ftrace=full. why?
AL> > AL> >> > example21 is special and requires as an argument the maximum number
AL> > AL> >> > of water molecules that you want to run (32-512). if you provide
AL> > AL> >> >
AL> > AL> >> > the error message stems from the numerical compares of the first
AL> > AL> >> > argument of the script, $1, with the corresponding number of waters.
AL> > AL> >> > the way it is done, allows to have a default value of '0' (i.e.
AL> > AL> >> > don't run any of those tests). your -ftrace=full argument expands
AL> > AL> >> > to a non-numerical value, hence the error messages. in all other
AL> > AL> >> > scripts it will be ignored, anyways.
AL> > AL> >> >
AL> > AL> >> > cheers,
AL> > AL> >> >    axel.
AL> > AL> >> >
AL> > AL> >> >
AL> > AL> >> > AL>
AL> > AL> >> > AL>    Amos Leffler
AL> > AL> >> > AL>
AL> > AL> >> > AL>     Walnut Creek, CA.
AL> > AL> >> > AL>
AL> > AL> >> > AL>        unaffiliated
AL> > AL> >> > AL>
AL> > AL> >> > AL>
AL> > AL> >> >
AL> > AL> >> >
AL> > AL> >>
AL> > AL> >> _______________________________________________
AL> > AL> >> Pw_forum mailing list
AL> > AL> >> Pw_forum at pwscf.org
AL> > AL> >> http://www.democritos.it/mailman/listinfo/pw_forum
AL> > AL> >>
AL> > AL> >>
AL> > AL> >
AL> > AL> >
AL> > AL> 
AL> > AL> 
AL> >
AL> >   
AL> 
AL> _______________________________________________
AL> Pw_forum mailing list
AL> Pw_forum at pwscf.org
AL> http://www.democritos.it/mailman/listinfo/pw_forum
AL> 

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.




More information about the users mailing list