R help - Sep 2009 - Cannot allocate memory of size x on Linux - what's the solution?

Hi all,

I'm running an analysis with the random forest tool. It's being applied
to a
data matrix of ~60,000 rows and between about 40 and 200 columns. I get the
same error with all of the data files (Cannot allocate vector of size
428.5MB). 

I found dozens of threads regarding this problem, but they never seem to be
concluded. Usually the OP is directed to the memory allocation help file
(which I haven't understood the solution for linux), and the last post is
the OP saying they haven't sorted out their problem yet. 

I'm running on a LINUX machine wtih 64GB RAM, so it's not a problem with
lack of system resources. 

Can anyone tell me how I can get R to allocate larger vectors on Linux? 

Many thanks,

Dave
-- 
View this message in context:
http://www.nabble.com/Cannot-allocate-memory-of-size-x-on-Linux---what%27s-the-solution--tp25659271p25659271.html
Sent from the R help mailing list archive at Nabble.com.

davew0000 wrote:
> Hi all,
> 
> I'm running an analysis with the random forest tool. It's being
applied to a
> data matrix of ~60,000 rows and between about 40 and 200 columns. I get the
> same error with all of the data files (Cannot allocate vector of size
> 428.5MB). 
> 
> I found dozens of threads regarding this problem, but they never seem to be
> concluded. Usually the OP is directed to the memory allocation help file
> (which I haven't understood the solution for linux), and the last post
is
> the OP saying they haven't sorted out their problem yet. 
> 
> I'm running on a LINUX machine wtih 64GB RAM, so it's not a problem
with
> lack of system resources. 
> 
> Can anyone tell me how I can get R to allocate larger vectors on Linux? 

1. Check how much memory R used at the point the error message appeared. 
If it is round about 60 Gb, you know that it is lack of resources - for 
the given problem. If it is much less (around 2Gb), you might have a 
32-bit R binary or you have some memory quota for your process.

Uwe Ligges


> Many thanks,
> 
> Dave