[R] Running R under Mosix
tlumley at u.washington.edu
Thu May 3 17:27:04 CEST 2001
On Thu, 3 May 2001 B.Rowlingson at lancaster.ac.uk wrote:
> Luke Tierney wrote:
> > Strange how these things go: we just had a discussion about this very
> > issue among the R core group yesterday.
> Possibly because I brought it up on the Mosix list and the
> discussion migrated across? :)
Yes, via Tony Rossini
> > However, Thomas Lumley, who has been using R with Mosix, mentioned
> > that there is a way you can tell Mosix that it is OK to migrate an R
> > process even though it makes many of these system calls--I'm sure he
> > can provide the details of how to do this.
> Yes, you can do 'mosrun -c R <args>', but you don't win. R still has
> to make the system calls and they slow down the processing. Unless
> got a spare Mosix node that is over twice as fast as your home node
> Mosix will keep it on the home node (with my test job, anyway).
It works for me. R processes that don't do much output migrate when I
start them with
cpujob R args at least after a short initial period while the memory
manager is getting itself sorted. Presumably it's sensitive to other setup
features, load, phase of the moon, but in my case the home node is
slightly faster than any other node.
In any case, the only currently known bad effect of changing the
SETJMP/LONGJMP macros to setjmp/longjmp is that on some systems (including
Debian) CTRL-C will work only once per session. For batch use this isn't
a problem, though for interactive use it is.
Thomas Lumley Asst. Professor, Biostatistics
tlumley at u.washington.edu University of Washington, Seattle
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch
More information about the R-help