R help - Jul 2004 - SMP, Single System Image clustering, multithreading

Hi All,

  I have looked through the following thread:

http://tolstoy.newcastle.edu.au/R/help/02b/4551.html

  and would like to ask after 1.5 year the same question:

  Are there any plans to involve multithread 
  computations in R?

Few word about my motivation:

  I use R on my OpenSSI cluster with 7 CPUs, so one 
  could simply think of this cluster as of single computer 
  with 7 CPU. My own C++ maths run good in parallel
  on the cluster. What is even better for me: my C++ code
  has *no* any statements about parallelization. 
  All parallezation is done at command line. When I 
  start a new piece of calculation from command line,
  the process is migrated to a free cluster node 
  automatically, it migrates even during the calculation.

  The whole project is calculated nearly up to 7 time 
  quicker then on a single CPU. Thus, I do see the boost 
  clearly.

  However the same approach applied to R will force me
  to use batch mode instead of nice interactive mode.

  Now I am writing this mail and R is still busy with
  stats in usual interactive mode. Other 6 processors
  are idling unfortunately. It would be really nice 
  to change this situation.

my best regards,
Valery.

I think there are plans to work on multithreading in R (see 
http://developer.r-project.org/) but it's probably a long-term 
project.  For now, you might want to try building R with a 
multithreaded BLAS.

-roger

Khamenia, Valery wrote:
> Hi All,
> 
>   I have looked through the following thread:
> 
> http://tolstoy.newcastle.edu.au/R/help/02b/4551.html
> 
>   and would like to ask after 1.5 year the same question:
> 
>   Are there any plans to involve multithread 
>   computations in R?
> 
> Few word about my motivation:
> 
>   I use R on my OpenSSI cluster with 7 CPUs, so one 
>   could simply think of this cluster as of single computer 
>   with 7 CPU. My own C++ maths run good in parallel
>   on the cluster. What is even better for me: my C++ code
>   has *no* any statements about parallelization. 
>   All parallezation is done at command line. When I 
>   start a new piece of calculation from command line,
>   the process is migrated to a free cluster node 
>   automatically, it migrates even during the calculation.
> 
>   The whole project is calculated nearly up to 7 time 
>   quicker then on a single CPU. Thus, I do see the boost 
>   clearly.
> 
>   However the same approach applied to R will force me
>   to use batch mode instead of nice interactive mode.
> 
>   Now I am writing this mail and R is still busy with
>   stats in usual interactive mode. Other 6 processors
>   are idling unfortunately. It would be really nice 
>   to change this situation.
> 
> my best regards,
> Valery.
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://www.stat.math.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html
>