University of Cambridge > Talks.cam > Microsoft Research Cambridge, public talks > Naiad: Iterative and Incremental Data-Parallelism using Differential Dataflow

Naiad: Iterative and Incremental Data-Parallelism using Differential Dataflow

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Microsoft Research Cambridge Talks Admins.

This event may be recorded and made available internally or externally via http://research.microsoft.com. Microsoft will own the copyright of any recordings made. If you do not wish to have your image/voice recorded please consider this before attending

Naiad is a new framework for data-parallel computing, based on potentially-cyclic dataflow graphs. Naiad’s execution model is based on processing successive differences between collections, which naturally enables efficient support for iterative algorithms and programs where the inputs change incrementally. In this talk, I will motivate and introduce our “differential dataflow” model, and describe the language (NaiadLINQ) and prototypical multithreaded implementation (NaiadMP) that we have developed. We implemented our current proof of concept for a single shared-memory machine, and I will present some preliminary results from this setting. We are currently investigating how the Naiad model can be extended to a cluster setting, and I will discuss some of the challenges and trade-offs that we are facing in that regard.

This is joint work with Frank McSherry, Rebecca Isaacs and Michael Isard.

This talk is part of the Microsoft Research Cambridge, public talks series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

© 2006-2017 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity