1 / 12

Presto: Distributed R for Big Data Power Method with Netflix ALS, 20x Faster

Learn how Presto, a distributed R system, utilizes the Power method with 1B edges, scaling Netflix ALS 20 times faster than In-memory Hadoop. See a speed demo, lj_matrix function, and darray operations for efficient big data processing. For inquiries, visit tinyurl.com/presto-project or email presto-dev@external.groups.hp.com.

kelii
Download Presentation

Presto: Distributed R for Big Data Power Method with Netflix ALS, 20x Faster

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Distributed R for big data Shivaram Venkataraman*, Indrajit Roy+, Alvin AuYoung+, Rob Schreiber+, Erik Bodzsar#, Kyungyong Lee^+ *UC Berkeley, +HP Labs, #U Chicago, ^ UFL

  2. Single Threaded + Single Machine R

  3. R R R R R

  4. darray

  5. foreach f (x)

  6. Power method with 1B edges, Netflix ALS Scale 20x faster than In-memory Hadoop Speed

  7. demo

  8. lj_matrixdarray(dim=c(n,n),blocks=c(n,n)) in_vectordarray(dim=c(n,1), blocks=(s,1), data=1/n) out_vector darray(dim=c(n,1), blocks=(s,1)) foreach(i, 1:length(splits(lj_matrix)), function(g = splits(lj_matrix, i), i = splits(in_vector), o = splits(out_vector, i)) { n  g %*% o update(n) })

  9. Contact us - alpha version tinyurl.com/presto-project hpl.hp.com/research/presto.htm presto-dev@external.groups.hp.com

  10. R R R R

More Related