[R] question about running out of memory on R -- memory.limit change in R 2.6.0?

[kees]

Earl,

Reported memory sizes work ok on Vista 64. If I ask for 3.5M it will give  
it, even though I have only 2Mb of RAM.
It devaults to 2Mb as expected.

Op Tue, 06 Nov 2007 19:38:07 +0100 schreef Earl F. Glynn  
<efg at stowers-institute.org>:

> "jim holtman" <jholtman at gmail.com> wrote in message
> news:644e1f320711051542s51a7e793t8cbc1f8e7cc35b at mail.gmail.com...
>> --max-mem-size=N
>> (Windows only) Specify a limit for the amount of memory to be used
>> both for R objects and working areas. This is set by default to the
>> smaller of 1.5Gb24 and the amount of physical RAM in the machine, and
>> must be between 32Mb and 3Gb.
>
> Something seems to have changed in R 2.6.0 -- the limit is no longer 3  
> GB.
>
> On a PC with 3.5 GB with R 2.5.1 using default command line:
>
> Default
>
>> memory.limit()
>
> [1] 1610612736
>
>
>
> With --max-mem-size=3000M on the command line:
>
>
>
>> memory.limit()
>
> [1] 3145728000
>
>
>
>
>
>
> Let's see what happens with R 2.6.0 on the same 3.5 GB PC (Windows XP):
>
>
>
> Default command line:
>
>
>> memory.limit()
>
> [1] 1535.875
>
>
>
> Note:  The format of the return value is now in MB instead of bytes.
>
>
>
> With --max-mem-size=3000M on the command line:
>
>
>
> Error message:  WARNING: --max-mem-size=3000M: too large and taken as  
> 2047M
>
>
>
>> memory.limit()
>
> [1] 2047.875
>
>
>
> This value is less than the value shown in R 2.5.1 (after converting  
> both to
> the same basis).
>
>
>
> But was this change a "fix" to report that a Windows process only has a  
> 2 GB
> address space?  That is, the value returned by R 2.5.1 of 3145728000  
> wasn't
> strictly correct?
>
>
>
> efg
>
>
>
> Earl F. Glynn
>
> Scientific Programmer
>
> Stowers Institute
>
> ______________________________________________
> 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.
>
>



-- 
Gemaakt met Opera's revolutionaire e-mailprogramma:  
http://www.opera.com/mail/