Making Large Legacy Software Run Twice as Fast on a Quad-Core with Just 1 Month of Programmer Effort

Subscribers:
344,000
Published on ● Video Link: https://www.youtube.com/watch?v=iJRSQmFzUgM



Duration: 1:07:58
79 views
2


A major challenge of the industry-wide shift to multicore processing is what to do with large legacy software that was not written with parallelism in mind. Ideally, the transition from sequential to parallel code would involve zero programmer effort: e.g., it would be handled automatically by a parallelizing compiler. In practice, however, this ideal remains beyond the scope of today's compiler technology for most applications. For complex applications that were developed by large teams over many years, the cost of rewriting key parts of the software nearly from scratch to introduce parallelism is prohibitively expensive. Fortunately, there is a secret




Other Videos By Microsoft Research


2016-09-06Putting our digital information in its place
2016-09-06Integrating art with information visualization to create interactive tabletop applications
2016-09-06Ecological Intelligence: How Knowing the Hidden Impacts of What We Buy Can Change Everything
2016-09-06Organizing Flickr: Object Mining using Particular Object Retrieval
2016-09-06Augmented Social Cogniton [1/12]
2016-09-06Automatically Finding Patches Using Genetic Programming
2016-09-06Transforming Software Powerhouse to Knowledge Systems Powerhouse
2016-09-06Your Money or Your Life: Nine Steps to Transforming Your Relationship with Money
2016-09-06Scheduling and Synchronization for Multicores
2016-09-06Developing Game-Themed Applications With XNA Game Studio: Session 2
2016-09-06Making Large Legacy Software Run Twice as Fast on a Quad-Core with Just 1 Month of Programmer Effort
2016-09-06(4,4)-split Jacobians of curves of genus 2
2016-09-06Advances in the CM method for elliptic curves
2016-09-06Normalisation by Evaluation
2016-09-06Utilising the Ubiquity of the Cell Phone to Record Physiological Activities
2016-09-06Distributed Storage Systems Made Easy
2016-09-06Software Model Checking for Confidentiality
2016-09-06Out of Our Heads: Why You Are Not Your Brain and Other Lessons from the Biology of Consciousness
2016-09-06Bill Gates, Warren Buffett and Other Great Minds Show how Creative Capitalism Can Save the World
2016-09-06Bayesian infinite matrix factorization
2016-09-06Securing the Web Platform



Tags:
microsoft research