Stencil Computation Auto-tuning on Modern Multicore Architectures

Subscribers:
344,000
Published on ● Video Link: https://www.youtube.com/watch?v=eNN0nGt772I



Duration: 1:12:24
474 views
10


Understanding the most efficient design and utilization of emerging multicore systems is one of the most challenging questions faced by the mainstream and scientific computing industries in several decades. Our work explores multicore stencil (nearest-neighbor) computations- a class of algorithms at the heart of many structured grid codes, including PDE solvers. We develop a number of effective optimization strategies in order to build an auto-tuning environment that minimizes runtime while maximizing performance portability. Within this auto-tuning framework, we also employ techniques like machine learning to find near-optimal parameter configurations quickly. To evaluate the effectiveness of these strategies, we explored the broadest set of multicore architectures in the current HPC literature, including the Intel Clovertown, AMD Barcelona, Sun Victoria Falls, IBM QS22 PowerXCell 8i, and NVIDIA GTX280. Overall, our auto-tuning optimization methodology results in the fastest multicore stencil performance to date. Finally, we present several key insights into the architectural trade-offs of emerging multicore designs and their implications on scientific algorithm development.




Other Videos By Microsoft Research


2016-09-07Audio Cameras for Audio-Visual Scene Analysis
2016-09-07Block Switching: Towards a Robust Protocol Stack for Diverse Wireless Networks
2016-09-07A Programming Language for the New Web
2016-09-07The Beauty and the Beast: Vulnerability in Red Hat's Packages
2016-09-07Debian: Anatomy of An Open Source Project
2016-09-07UPCRC Multicore Applications Workshop - Session # 3 - Social Interaction
2016-09-07Supersingular abelian varieties and modular forms
2016-09-07The Jasons: The Secret History of Science's Postwar Elite           
2016-09-07UPCRC Multicore Applications Workshop - Session # 4 - Speech and Audio
2016-09-07Literacy Bridge and the Talking Book Project
2016-09-07Stencil Computation Auto-tuning on Modern Multicore Architectures
2016-09-07MSPAC Discussion and Book Signing with Senator John Kerry and Teresa Heinz Kerry
2016-09-07Mark-Region and Other Advances in Garbage Collection
2016-09-07The Medea Hypothesis: Is Life on Earth Ultimately Self Destructive?
2016-09-07WaveScope: Stream Programming on Heterogeneous Wireless Devices
2016-09-07Bags of words: the search engine
2016-09-07Intrinsic Robustness of the Price of Anarchy
2016-09-07Reach for A*: an Efficient Point-to-Point Shortest Path Algorithm
2016-09-07ABC-MART: Recent Improvements in Boosting, Trees and Classification Algorithms
2016-09-07Intelligent Fault Notification through Understanding Developer Behavior
2016-09-07Contextual Link Analysis for Web Search



Tags:
microsoft research