[BioC] memory error on unix
Kasper Daniel Hansen
khansen at stat.Berkeley.EDU
Wed Jun 13 22:29:48 CEST 2007
On Jun 13, 2007, at 10:52 AM, Brandon Higgs wrote:
> I updated my mapping files (nsp and sty) to 0.2.6 and oligo to
> version 1.1 and on a 4GB (32 bit) machine with increased swap to
> ridiculous amounts (30GB) and I still am unable to create a
> snpqset. I can read a single cel in, but at snprma, I get stuck.
> Can't wait for that dev version of oligo Benilton! Please
> distribute the path when you get it done. Many thanks for your
> hard work!
On a 32 bit machine, a single process cannot use more than x GB (with
x depending a bit on the OS, but typically something like 2-3 GB), so
the huge swap does not help you at all.
You might want to look for aroma.affymetrix when Henrik Bengstsson
gets around to releasing that (in August he says). That will also
help you process an unlimited number of snp arrays in bounded memory
(typically something like 500mb). Of course, only certain algorithms
are implemented in that package.
> The memory error just means that you ran out of memory, it does not
> mean that you have less than 1.8GB in total. If you are on a 32bit OS
> you will not be able to use all 8 GB of ram. That has nothing to do
> with R per se.
> On Jun 11, 2007, at 11:03 AM, Beth Wilmot wrote:
>> I am running R-2.5.0 and Bioconductor 2.0 and oligo 1.0.2-3 on a Sun
>> OS 5.10 Generic_118855-19 i86pc i386 i86pc and Solaris 10 6/06
>> s10x_u2wos_09a X86.
>> I have seen the help messages concerning memory errors but cannot
>> a solution to unix memory errors. I thought R would be using all
>> available memory?
>> I have 8Gb RAM and 4Gb swap available.
>> I am running one 500K test chip and when I try to create a
>> SnpQSet in
>> oligo I get the following error:
>> Error: cannot allocate vector of size 1.8 Gb
>> $ ulimit -a
>> time(seconds) unlimited
>> file(blocks) unlimited
>> data(kbytes) unlimited
>> stack(kbytes) unlimited
>> coredump(blocks) unlimited
>> nofiles(descriptors) 256
>> memory(kbytes) unlimited
> that gives answers, not web links.
> [[alternative HTML version deleted]]
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> Search the archives: http://news.gmane.org/
More information about the Bioconductor