[R-br] Fwd: [Computing] byte-compiler and parallel computing in R

Leonard de Assis assis.leonard em gmail.com
Quinta Novembro 10 19:51:11 BRST 2011



-------- Mensagem original --------
Assunto: 	[Computing] byte-compiler and parallel computing in R
Data: 	Thu, 10 Nov 2011 12:37:06 -0800
De: 	Jan de Leeuw <deleeuw em stat.ucla.edu>
Para: 	statcompute <computing em stat.ucla.edu>



R-2.14.0 comes with Luke Tierney's byte-compiler installed. Byte-compilation can speed up R computations by
a factor 5, although if your functions are just calls to compiled C code the speedup will be much
smaller. By default, only the required packages are byte-compiled. But if you set (in your bash profile)

export R_COMPILE_PKGS=1

then all packages will byte-compile when installed.

If you set

export R_ENABLE_JIT=3

then all functions you define in your interactive or batch session will be byte-compiled the first time they
are entered into the workspace (and the same is true for all loops).

It should be noted that some contributed packages have namespace problems when byte-compiled.

R-2.15.0 (i.e. R-devel) has parallel versions of Map() and mapply(). parallel makes fork or socket
cluster and has parallel random number generation. MPI clusters are handled (for now) by passing the
information to the snow package.

===
Jan de Leeuw; Distinguished Professor and Chair, UCLA Department of Statistics;
Editor: Journal of Multivariate Analysis, Journal of Statistical Software;
US mail: 8125 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554
phone (310)-825-9550;  fax (310)-206-5658;  email: deleeuw em stat.ucla.edu
.mac: jdeleeuw ++++++  aim: deleeuwjan ++++++ skype: j_deleeuw
homepages: http://gifi.stat.ucla.edu ++++++ http://www.cuddyvalley.org
  -------------------------------------------------------------------------------------------------
           No matter where you go, there you are. --- Buckaroo Banzai
                    http://gifi.stat.ucla.edu/sounds/nomatter.au
  -------------------------------------------------------------------------------------------------




_______________________________________________
Computing mailing list
Computing em lists.stat.ucla.edu
http://lists.stat.ucla.edu/mailman/listinfo/computing


-------------- Próxima Parte ----------
Um anexo em HTML foi limpo...
URL: <http://listas.inf.ufpr.br/pipermail/r-br/attachments/20111110/f8ab4169/attachment.html>


Mais detalhes sobre a lista de discussão R-br