CODEWITHSUNDEEP
rparallel-processing
m0nhawk
m0nhawk

Reputation: 24168

Can't use all cores in R

I was trying to make a parallel loop, but I've found R can't utilize all my 8 cores, it uses only 1 with this code:

library(parallel)
library(foreach)
library(doParallel)
no_cores <- detectCores() - 1
cl <- makeCluster(no_cores)
registerDoParallel(cl, cores = no_cores)

sum.of.squares <- foreach(i = 1:10e4, .combine = "+", .init = 0) %dopar% {
  sqrt(i)
}

stopCluster(cl)

It uses only about 13-15% of CPU. Why?

Using Windows 8.1 with R 3.2.1.

Upvotes: 1

Views: 2288

Answers (1)

user3652621
user3652621

Reputation: 3634

Use R Revolution Open (RRO) 3.2

http://mran.revolutionanalytics.com/documents/rro/installation/

It comes integrated with the Intel MKL parallel math libraries.

See a fuller comparison here.

Upvotes: 1

Related Questions