QA: How do Revolution R Enterprise and RevoScaleR use multiple CPUs?

R is not inherently a multi-threaded application, so in normal circumstances it only uses one processor at a time. 

There are several options for parallel programming that provide value: 

1. Revolution R Enterprise will use all available processors for some common math operations, like matrix multiplication. (It is linked with multi-threaded math libraries which improve performance on multi-core Intel processors.) 

2. The RevoScaleR package, shipped with Revolution R Enterprise, offers parallel external memory algorithms and a very efficient data file format (.xdf). 

3. You can also write explicit parallel code using ParallelR. See CRAN Task View: High-Performance and Parallel Computing with R for a list of options: 

http://cran.r-project.org/web/views/HighPerformanceComputing.html 

4. The RevoScaleR package included with Revolution R Enterprise provides new tools for parallel and distributed computing with R that can scale out.   
Eigenschappen

Artikel-id: 3104264 - Laatst bijgewerkt: 29 okt. 2015 - Revisie: 1

Feedback