[R-sig-hpc] Problems parallelizing glmnet
Brian G. Peterson
brian at braverock.com
Fri Sep 7 02:31:47 CEST 2012
On 09/06/2012 04:09 PM, Peter Langfelder wrote:
> FWIW, the foreach construct itself (without any parallel backend) is
> quite slow and I would not use to loop over a large number of quick
> calculations.
Peter, this is right. It is not a general replacement for a for
loop.... there's too much infrastructure involved. However, we
typically write all of our code that will be parallelized for deployment
using %dopar% and then use registerDoSEQ() and run single threaded on a
smaller test set while debugging.
Regards,
- Brian
--
Brian G. Peterson
http://braverock.com/brian/
Ph: 773-459-4973
IM: bgpbraverock
More information about the R-sig-hpc
mailing list