Parallelizing R across multiple nodes

timolin

Hi,

 

I've been able to deploy some R functions using the parallel package and mclapply to take advantage of multiple cores within a single node. 

 

Has anyone deployed a R function through UDX that takes advantage of CPU cores across multiple nodes?  I was looking at the snow package to do this and was hoping from tips or at least some existence proof that someone had pulled this off.

 

Thanks, Tim 

SruthiA

Hi,

 

   If you want to achieve parallelism using R across all nodes, You can try using our new product "Distributed R". HP Distributed R is a high-performance scalable platform for the R language. It enables R to leverage multiple cores and multiple servers to perform Big Data Advanced Analytics. It consists of new R language constructs to easily parallelize algorithms across multiple R processes.

 

For more information on  our new product, please refer to the URL

 

http://my.vertica.com/docs/DISTR/1.1.x/HTML/index.htm#DistributedR/AboutDistributedR/About_Distributed_R.htm%3FTocPath%3DAbout%2520HP%2520Distributed%2520R%7C_____0

 

It is worth trying this product and I am sure you will be impressed.

 

-Regards,

 Sruthi