R Parallel computing: select which objects to be distributed into cores

Question

I have a question related to r-parallel computing. I am using something like:

cl.tmp = makeCluster(10, type="SOCK")
registerDoParallel(cl.tmp)
AA <- foreach(i = 1:48, .inorder = TRUE, .combine = rbind, .verbose=TRUE) %dopar% {
# A function that depends on some selected objects in current environment
}
stopCluster(cl.tmp)

How to identify some particular objects in the current environment to be distributed into processor cores and so be used for some functions? I do not need R to copy the whole objects in the environment into different cores, but only some selected objects. In my project, I have big R objects and I do not need to copy/distribute them into cores, to avoid ram problems. Is there a solution for that?

Thanks

Assuming that you know which objects are to be exported (that's not entirely clear from your question), you can insert something like `clusterExport(cl.tmp, c(obj1, obj2, ...), globalenv())` on line 2 — ChrKoenig, Aug 13 '15 at 19:38

score 0 · Answer 1 · edited May 23 '17 at 10:29

0

Take a look at this: reading global variables using foreach in R

Only variables referenced inside the foreach loop are copied from the global environment.

edited May 23 '17 at 10:29

Community

1
1

answered Aug 13 '15 at 18:48

mattdevlin

1,045
2
10
17

R Parallel computing: select which objects to be distributed into cores

1 Answers1