Carl Boettiger

Some notes for ref­er­ence on par­al­lel com­put­ing strate­gies on R with ded­i­cated high per­for­mance clus­ters. par­al­leliza­tion in R For clus­ter com­put­ing in R, I’ve found the “snow” pack­age with MPI to be the most flex­i­ble and robust solu­tion. snow­fall is a use­ful pack­age to use mul­ti­ple cores on a sin­gle proces­sor, but request­ing all 16 threads …

Can I get a generic mon­te­car­lotest func­tion work­ing for both warn­ingsig­nals and pmc? Needs to be able to han­dle S4 classes as well as S3 classes, or need to define the meth­ods update, sim­u­late, log­lik and get­Pa­ra­me­ters for any object that would use the method. i.e. these need to be defined in warn­ingsig­nals for its func­tions …

Parellelization/Scaling of code MPI on farm clus­ter Got my MPI codes run­ning. Much bet­ter way to get jobs into the queue, ask­ing for 16 threads that don’t have to be on the same node is much faster. Can also ask for 161 threads, but will wait longer in the queue. The trick to get­ting this to run …

After a few help­ful emails from pack­age cre­ator JD Long, I have the segue pack­age run­ning with cus­tom R pack­ages. The pack­age is avail­able on Google code. With two lines of code I can start sub­mit­ting jobs to very large clus­ters of com­put­ers on the Ama­zon cloud. For a basic intro­duc­tion to the pack­age see …