Skip to content

Memory management

3 messages · Konstantinos Euripides Vorloou, Peter Dalgaard, Ross Ihaka

Konstantinos Euripides Vorloou
# 
Dear all,

I don't get it:
First of all, the help doesn't say what are the memory limits of
R. Say, what's the max heap size for instance ????

Secondly,  I invoke R with the following commands each time:

rgui --vsize 30M --nsize 1000K
rgui --vsize 30M --nsize 2000K
rgui --vsize 30M --nsize 3000K
rgui --vsize 30M --nsize 4000K

I try to open a matrix 8000x8000 by issuing

x<-matrix(nrow=8000,ncol=8000)

I  always get the following message:

Error: heap memory (20480  Kb) exhausted [needed 250000 Kb more]

It's always 250000 Kb less. Although gc() reports the memory management
as
expected (correctly ?), increased every time I invoke R with greater
vsize,nsize
params. My matrix should hold roughly 1 Meg of memory.
Is this correct ?

When I issue :

rgui --vsize 30M --nsize 5000K

R hangs up and have to kill the session.

I am using R on a 333 Celeron Pentium PC with 128Megs of memory
under windows 98. I don't think I should have a problem there. Hardware
seems ok...

Would the interpreter normally use the swap disk to
create a ramdrive for extra memory needs ? When I use
large vsize, nsize params there is an awful lot of disk swapping...
This is more obvious on slower pentiums.

I'd hate to recompile R to enlarge the heap. Is there another
way out of it ? Anything undocumented ?

Thanks for your time and attention.

Best regards,
Costas

--
-------------------------------------------------------------------------

Konstantinos E. Vorloou           | Tel: +44 (0)191 374 1821
Department of Economics & Finance | Fax: +44 (0)191 374 7289
University of Durham,             | email: K.E.Vorloou at durham.ac.uk
23/26 Old Elvet,                  | or : vorlow at rocketmail.com
Durham DH1 3HY,                   | or : Vorlow at bigfoot.com
UK.                               | Express:
20595534 at pager.mirabilis.com
                                  | ICQ number: 20595534
http://www.durham.ac.uk/Economics | http://www.durham.ac.uk/~dec3kev
-------------------------------------------------------------------------



-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Peter Dalgaard
# 
Konstantinos Euripides Vorloou <K.E.Vorloou at durham.ac.uk> writes:
it's --vsize that controls the heap...
I get 8000*8000*8 = 512000000 bytes, alias 488Meg...
Ross Ihaka
#
On 27 Apr 1999, Peter Dalgaard BSA wrote:

            

      
      
      
    

        

        
On 27 Apr 1999, Peter Dalgaard BSA wrote:

        

            

      
      
      
    

        
Peter's arithmetic is correct.  By default R stores numeric values
in double precision (32-bit integers are the other possibility).

There are 8000*8000 elements in the matrix, each occupying 8 bytes
(the size of a double).

This is a very large matrix.  R isn't designed to handle problems on this
scale (not many interactive systems are).  If you really want to try
problems of this scale, you'd be better off trying Fortran or C.

	Ross

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._