derek eder
2010-Nov-23 08:26 UTC
[R] Error: cannot allocate vector of size x Gb (64-bit ... yet again)
Hello,
I am facing the dreaded "Error: cannot allocate vector of size x Gb"
and
don't understand
enough about R (or operating system) memory management to diagnose and
solve the problem
-- despite studying previous posts and relevant R help -- e.g.:
"Error messages beginning cannot allocate vector of size indicate a
failure to obtain memory,
either because the size exceeded the address-space limit for a process
or, more likely,
because the system was unable to provide the memory.
[...] On all builds of R, the maximum length (number of elements)
of a vector is 2^31 - 1 ~ 2*10^9, as lengths are stored as signed integers.
In addition, the storage space cannot exceed the address limit."
- from Memory-limits {Base}
Simple question: Given 64-bit R (AMD64 Linux) with a ulimit of
"unlimited", can the size of an R object exceed the amount of
availlable RAM
memory?
Empirically my system with 4Gb RAM and ample Swap, is failing:
> x <- integer(10^9)
> object.size(x)
4000000040 bytes
> gc()
used (Mb) gc trigger (Mb) max used (Mb)
Ncells 121195 6.5 350000 18.7 350000 18.7
Vcells 500124024 3815.7 606849099 4629.9 550124408 4197.2
> matrix(x, ncol=16)
Error: cannot allocate vector of size 3.7 Gb
I don't understand how this operation violates the limits detailed in
the Memory-limit help (above).
Thank you!
Derek Eder
-------------------------------------------------------------------------------------------------
> version
_
platform x86_64-pc-linux-gnu
arch x86_64
os linux-gnu
system x86_64, linux-gnu
status
major 2
minor 11.1
year 2010
month 05
day 31
svn rev 52157
language R
version.string R version 2.11.1 (2010-05-31)
derek at papanca:~$ top
top - 09:10:18 up 51 min, 4 users, load average: 0.51, 0.51, 0.45
Tasks: 160 total, 2 running, 158 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.0%us, 25.0%sy, 0.0%ni, 75.0%id, 0.0%wa, 0.0%hi, 0.0%si,
0.0%st
Mem: 3796484k total, 3764852k used, 31632k free, 14204k buffers
Swap: 2929660k total, 834240k used, 2095420k free, 94800k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
2854 derek 20 0 239m 9260 5448 S 6 0.2 0:05.53
gnome-terminal
1164 root 20 0 218m 31m 10m S 4 0.8 1:29.71 Xorg
3331 derek 20 0 19276 1324 944 R 1 0.0 0:00.6 top
Uwe Ligges
2010-Nov-23 11:17 UTC
[R] Error: cannot allocate vector of size x Gb (64-bit ... yet again)
On 23.11.2010 09:26, derek eder wrote:> Hello, > > I am facing the dreaded "Error: cannot allocate vector of size x Gb" and > don't understand > enough about R (or operating system) memory management to diagnose and > solve the problem > -- despite studying previous posts and relevant R help -- e.g.: > > "Error messages beginning cannot allocate vector of size indicate a > failure to obtain memory, > either because the size exceeded the address-space limit for a process > or, more likely, > because the system was unable to provide the memory. > [...] On all builds of R, the maximum length (number of elements) > of a vector is 2^31 - 1 ~ 2*10^9, as lengths are stored as signed integers. > In addition, the storage space cannot exceed the address limit." > - from Memory-limits {Base} > > > Simple question: Given 64-bit R (AMD64 Linux) with a ulimit of > "unlimited", can the size of an R object exceed the amount of availlable > RAM > memory? > > Empirically my system with 4Gb RAM and ample Swap, is failing: > > > x <- integer(10^9) > > > object.size(x) > 4000000040 bytes > > > gc() > used (Mb) gc trigger (Mb) max used (Mb) > Ncells 121195 6.5 350000 18.7 350000 18.7 > Vcells 500124024 3815.7 606849099 4629.9 550124408 4197.2 > > > matrix(x, ncol=16) > Error: cannot allocate vector of size 3.7 Gb > > I don't understand how this operation violates the limits detailed in > the Memory-limit help (above).Why? If you have a limit of 4Gb, 4Gb are already used and you try to work with a 4Gb objeczt in the matrix function which makes a copy and creates another copy (with dim attributed), so 8Gb additionally. You will need at least 12 of your 4Gb. Probably R does not get a sufficient amount of non-fragmented memory from the OS or your swap space is not "ample". To make the above more efficient, just use: dim(x) <- c(10^8/16, 16) and you won't get any copies. Uwe> Thank you! > > > Derek Eder > > > > ------------------------------------------------------------------------------------------------- > > > > version > _ > platform x86_64-pc-linux-gnu > arch x86_64 > os linux-gnu > system x86_64, linux-gnu > status > major 2 > minor 11.1 > year 2010 > month 05 > day 31 > svn rev 52157 > language R > version.string R version 2.11.1 (2010-05-31) > > > > derek at papanca:~$ top > > top - 09:10:18 up 51 min, 4 users, load average: 0.51, 0.51, 0.45 > Tasks: 160 total, 2 running, 158 sleeping, 0 stopped, 0 zombie > Cpu(s): 0.0%us, 25.0%sy, 0.0%ni, 75.0%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st > Mem: 3796484k total, 3764852k used, 31632k free, 14204k buffers > Swap: 2929660k total, 834240k used, 2095420k free, 94800k cached > > PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND > 2854 derek 20 0 239m 9260 5448 S 6 0.2 0:05.53 gnome-terminal > 1164 root 20 0 218m 31m 10m S 4 0.8 1:29.71 Xorg > 3331 derek 20 0 19276 1324 944 R 1 0.0 0:00.6 top > > ______________________________________________ > R-help at r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide > http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code.
Prof Brian Ripley
2010-Nov-24 09:02 UTC
[R] Error: cannot allocate vector of size x Gb (64-bit ... yet again)
On Tue, 23 Nov 2010, derek eder wrote:> Hello, > > I am facing the dreaded "Error: cannot allocate vector of size x Gb" and > don't understand > enough about R (or operating system) memory management to diagnose and solve > the problem > -- despite studying previous posts and relevant R help -- e.g.: > > "Error messages beginning cannot allocate vector of size indicate a failure > to obtain memory, > either because the size exceeded the address-space limit for a process or, > more likely, > because the system was unable to provide the memory. > [...] On all builds of R, the maximum length (number of elements) > of a vector is 2^31 - 1 ~ 2*10^9, as lengths are stored as signed integers. > In addition, the storage space cannot exceed the address limit." > - from Memory-limits {Base} > > > Simple question: Given 64-bit R (AMD64 Linux) with a ulimit of "unlimited", > can the size of an R object exceed the amount of availlable RAM > memory? > > Empirically my system with 4Gb RAM and ample Swap, is failing: > >> x <- integer(10^9) > >> object.size(x) > 4000000040 bytes > >> gc() > used (Mb) gc trigger (Mb) max used (Mb) > Ncells 121195 6.5 350000 18.7 350000 18.7 > Vcells 500124024 3815.7 606849099 4629.9 550124408 4197.2 > >> matrix(x, ncol=16) > Error: cannot allocate vector of size 3.7 Gb > > I don't understand how this operation violates the limits detailed in the > Memory-limit help (above).It doesn't, and that example works on my (32GB RAM) 64-bit system> gc()used (Mb) gc trigger (Mb) max used (Mb) Ncells 134054 7.2 350000 18.7 350000 18.7 Vcells 1000131158 7630.4 1575382823 12019.3 1500131550 11445.1 So maybe you don't have 'ample swap' and/or have a VM limit set. Note that dim(x) <- c(length(x)/16, 16) is much more efficient use of memory.> > Thank you! > > > Derek Eder > > > > ------------------------------------------------------------------------------------------------- > >> version > _ > platform x86_64-pc-linux-gnu > arch x86_64 > os linux-gnu > system x86_64, linux-gnu > status > major 2 > minor 11.1 > year 2010 > month 05 > day 31 > svn rev 52157 > language R > version.string R version 2.11.1 (2010-05-31) > > > > derek at papanca:~$ top > > top - 09:10:18 up 51 min, 4 users, load average: 0.51, 0.51, 0.45 > Tasks: 160 total, 2 running, 158 sleeping, 0 stopped, 0 zombie > Cpu(s): 0.0%us, 25.0%sy, 0.0%ni, 75.0%id, 0.0%wa, 0.0%hi, 0.0%si, > 0.0%st > Mem: 3796484k total, 3764852k used, 31632k free, 14204k buffers > Swap: 2929660k total, 834240k used, 2095420k free, 94800k cached > > PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND > 2854 derek 20 0 239m 9260 5448 S 6 0.2 0:05.53 gnome-terminal > 1164 root 20 0 218m 31m 10m S 4 0.8 1:29.71 Xorg > 3331 derek 20 0 19276 1324 944 R 1 0.0 0:00.6 top > > ______________________________________________ > R-help at r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >-- 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