[CPMD-list] MEMORY ALLOCATION

[Axel Kohlmeyer]

>>> "VVS" == Veronique Van Speybroeck <Veronique.VanSpeybroeck at rug.ac.be> writes:

VVS> On Wed, 11 Sep 2002, Axel Kohlmeyer wrote:

>> 
>> >>> "VVS" == Veronique Van Speybroeck <Veronique.VanSpeybroeck at rug.ac.be> writes:
>> 
VVS> Hi,
>> 
VVS> I have a problem running a job that requires a large amount of memory.
VVS> I get something like this :
VVS> ***    MEMORY| THE NEW SIZE OF THE PROGRAM IS 1876066 kBYTES ***
>> 
VVS> ================================================================
VVS> BIG MEMORY ALLOCATIONS
VVS> PME            100521120                C2              20104216
VVS> C0              20104216                YLMB            15729183
VVS> YF              14378114                XF              14378114
VVS> TWNL             8740960                RHOPS            8738435
VVS> VPS              8738435                QRAD             6650100
VVS> ----------------------------------------------------------------
VVS> [PEAK NUMBER   88]      PEAK MEMORY    230758164 = 1846.1 MBytes
VVS> ================================================================
>> 
>> 
VVS> PROGRAM STOPS IN SUBROUTINE MEMORY| ALLOCATION FAILED (GDE)
>> 
VVS> Although my machine has 4GB RAM memory.  How can I solve this problem?
>> 
>> hello veronique,
>> 
>> it would help a lot if you would tell us, _what_ type of machine you
>> have (i.e cpu type, operating system, etc).
>> 
>> cheers,
>> axel.
>> 
VVS> Thanks
VVS> Veronique
>> 
VVS> 
VVS> -----------------------------------------------------------------------
VVS> Dr. ir. Van Speybroeck Veronique
VVS> Laboratorium voor Theoretische Fysica
VVS> Universiteit Gent
VVS> Proeftuinstraat 86
VVS> 9000 Gent
VVS> Tel +32-9-264.65.58  GSM : +32/474/259767
VVS> Fax +32-9-264.65.60
VVS> email : veronique.vanspeybroeck at rug.ac.be
VVS> http://inwfaxp2.rug.ac.be/~web/onderzoek/proj_3.html
VVS> -----------------------------------------------------------------------
>> 
VVS> _______________________________________________
VVS> CPMD-list mailing list
VVS> CPMD-list at cpmd.org
VVS> http://www.cpmd.org/mailman/listinfo/cpmd-list
>> 

VVS> I have tried the calculation on an Origin 2000 SGI machine with 4
VVS> processors and 4 GB RAM memory.  Operating system IRIX 6.5

VVS> I also tried the calculation on a PC with linux redhat 7.2, with 2 GB RAM
VVS> memory and two processors.

ok, 

VVS> ***    MEMORY| THE NEW SIZE OF THE PROGRAM IS 1876066 kBYTES ***

this means you already have successfully allocated nearly 2 gigabyte 
of memory and the _next_ allocation is failing.

on a pc, since it is a 32-bit cpu and due to limitations of its memory
management unit, you cannot (easily) have more than 2GB of (virtual)
memory per process (although you may have more memory+swap in total that
are used).

on the SGI you have to be careful to compile cpmd in 64-bit mode. 
otherwise the same concerns as with the pc apply.

furthermore you need an appropriate amount of swap to back up all
physical memory. if you do not have enough physical swap you may get
away with configuring 'virtual swap'. 

to explain: in it's default configuration the irix 6.x operating system
_requires_ to have enough swap available to cover _all_ allocated
physical memory, even for shared memory mappings so that you might under
some circumstances need much more than twice the amount of physical
memory as swap to simply use all the physical memory. with virtual swap
you are basically 'faking' this swap, which is perfectly ok in most cases.

hope this helps,
     axel.


VVS> Regards

VVS> veronique
>> 
>> 
>> --
>> 
>> =======================================================================
>> 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.
>> 


--

=======================================================================
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.