[R] How to increase memory for R on Soliars 10 with 16GB and 64bit R

Dongseok Choi choid at ohsu.edu
Wed Jul 13 20:34:54 CEST 2005


Thank you very much for your help!!
Now, it runs without any problem.

Is it going to be fixed in the next release?

Thanks again,
Dongseok




Dongseok Choi, Ph.D.
Assistant Professor
Division of Biostatistics
Department of Public Health & Preventive Medicine
Oregon Health & Science University
3181 SW Sam Jackson Park Road, CB-669
Portland, OR 97239-3098
TEL) 503-494-5336
FAX) 503-494-4981
choid at ohsu.edu

>>> "Prof Brian Ripley" <ripley at stats.ox.ac.uk> 07/13/05 12:03 AM >>>
On Tue, 12 Jul 2005, Dongseok Choi wrote:

>  My machine is SUN Java Workstation 2100 with 2 AMD Opteron CPUs and 16GB RAM.
>  R is compiled as 64bit by using SUN compilers.
>  I trying to fit quantile smoothing on my data and I got an message as below.
>
>> fit1<-rqss(z1~qss(cbind(x,y),lambda=la1),tau=t1)
> Error in as.matrix.csr(diag(n)) :
cannot allocate memory block of size 2496135168
>
>  The lengths of vector x and y are both 17664.
>  I tried and found that the same command ran with x[1:16008] and y[1:16008].
>  So, it looks to me a memory related problem, but I'm not sure how I can allocate memory block.
>   I read the command line option but not sure what do to with it.
>   Could you help me on this?

It is trying to allocate a single memory block of size over 2^31-1 bytes. 
R internally uses ints for sizes of vectors and that is a limit (see 
help("Memory-limits") ).  However, it is intended that on 64-bit systems 
that there is a limit here of 8*(2^31-1) but there was a typo.  Please 
change line 1534 of src/main/memory.c to

#if SIZEOF_LONG > 4

and re-compile.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk 
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/ 
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595




More information about the R-help mailing list