r - Maximum number of parallel processors in clusterR function? -
i using clusterr()
function predict randomforest
model on huge data set. therefore use cluster 100+ cores. code looks this:
library(randomforest) library(raster) library(parallel) r <- stack("...some/multilayer/raster...") rfo <- # load calculated randomforest object begincluster(100) rpred <- clusterr(r, predict, args = list(model = rfo))
i expected 100 initialized processors used. however, when @ top
in bash, 20 - 30 processes running @ full speed. others on sleep.
when getcluster()
cluster 100 sockets running.
is there limitation maximum number of processors in clusterr
function or parallel
backend?
is there possibility determine maximum number of processors used in clusterr
?
Comments
Post a Comment