[R] Parallel computing on Windows (foreach)

R. Michael Weylandt michael.weylandt at gmail.com
Thu Jul 26 18:32:26 CEST 2012


It's sad, but not an impossible result with synchronization overhead
(though I wouldn't have guessed it would be that bad) -- can you try
it on a more reasonable benchmark?

Also, that advice might be somewhat out of date -- why not use the
tools provided in

 library(parallel)

available for recent versions of R?

Michael

On Thu, Jul 26, 2012 at 8:00 AM, deep123 <deps_123 at yahoo.co.in> wrote:
> Hello all,
> My problem is similar to Sergey's.
> I could not try what Mario suggested, since doMPI is not available for
> windows.
>
> I have tried the Cedrick's commands, i.e.
> cl.tmp = makeCluster(rep("localhost",2), type="SOCK")
> registerDoSNOW(cl.tmp)
>
> but this is instead increasing the time taken. For Sergy's set of commands,
> i get,
>> x <- numeric(10000)
>> system.time(for(i in 1:10000) x[i] <- sqrt(i))
>    user  system elapsed
>    0.03    0.00    0.03
>> system.time(system.time(x <- foreach(i=1:10000, .combine="c") %do%
>> sqrt(i)))
>    user  system elapsed
>    4.71    0.00    4.71
>> system.time(system.time(x <- foreach(i=1:10000, .combine="c") %dopar%
>> sqrt(i)))
>    user  system elapsed
>    7.35    1.07  127.06
>
> i.e. %dopar% is taking longer than %do%
>
> Please guide me.
>
>
>
> --
> View this message in context: http://r.789695.n4.nabble.com/Parallel-computing-on-Windows-foreach-tp2256082p4637925.html
> Sent from the R help mailing list archive at Nabble.com.
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> 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.



More information about the R-help mailing list