The use of High Performance Computing continues to grow in the enterprise and beyond. In this podcast, James Reinders from Intel describes how Intel will continue to drive HPC democratization in 2016.
“At Intel, our passion to help drive the democratization of HPC is exemplified by many things. Here is my list of ten things which caught my attention as being most significant as we enter 2016, in no particular order:
- Code Modernization: Intel helps make “code modernization” more than just a hot topic by providing events and training (attended by tens of thousands of developers around the world in this past year), and many valuable resources, including online training, on the web. The key to me is how this helps us all “Think Parallel”. This was reinforced to me when, as part of our Code Modernization work in 2015, we ran a student competition. When the grand prize winner, who had reduced the application run time from 45 hours to under 9 minutes, was asked what the greatest tool he utilized, he said ‘my brain.’ I hope, by our code modernization efforts, to train more of our brains to “Think Parallel”. This is an important dialogue that will continue strongly in 2016.
- OpenHPC: Intel is a founding member and makes significant contributions to the OpenHPC community. OpenHPC is driven by a desire to aggregate a number of common ingredients required to deploy and manage High Performance Computing (HPC) Linux clusters. This in turn, makes deploying an HPC system easier. I’ve enjoyed the early debates among community members as to the best solutions for OpenHPC to embrace. The debates will help us all do better, and I look forward to seeing what 2016 brings as the community forges ahead and grows.
- Pearls Books: Learning by examples is critical in order to build expertise in any subject. When the examples come from world experts who share the challenges and successes of effectively using parallel programming, the learning available is substantial. I enjoyed editing these volumes both because of the experts with whom I was privileged to work, and the lessons I learned as well. High Performance Parallelism Pearls Volumes 1 and 2 are definitely worth reading to help in our drive to “Think Parallel”.
- Free Libraries for All: Yes, you can get four of the best libraries anywhere for free, no strings attached. Intel Math Kernel Library, Intel Data Analytics Library, Intel Integrated Performance Primitives, and Intel Threading Building Blocks can all be obtained at a single website for community licenses from Intel. The libraries are also part of Intel Parallel Studio XE 2016 as well, so you can choose free with community support or purchased with direct-from-Intel support. Either way, these libraries are very powerful.
- Free Tools for instructors, classroom, academic researchers and open source work. Intel’s award winning tools lead the industry in support for high performance programming. They are worth the price. But, if youqualify for free access then you have no excuse to not be using these incredibly powerful libraries.
- Big Data and Machine Learning: The Intel Data Analytics Acceleration Library (DAAL) premiered in 2015 and has proven its value in accelerating Big Data processing including Machine Learning. Most Big Data and Machine Learning work is done on Intel-based machines, and Intel DAAL gives that work an additional boost.
- Intel Parallel Studio XE 2016: With new features like Vectorization Advisor, MPI Snapshot and Intel Data Analytics Acceleration Library (DAAL), Intel’s industry leading tools continue to give software developers the best capabilities to optimize performance for Intel architecture.
- Second Generation Intel Xeon Phi Processors: With our first three systems deployed outside Intel using pre-production “Knights Landing” processors in 2015, the excitement and anticipation for the Second Generation of Intel Xeon Phi processors has never been higher. The revolutionary move to put Intel’s many-core architecture into a processor will help make the high levels of performance available to an unprecedented range of users in 2016. My personal goal to help this is to publish a useful book about it, with my colleagues at Intel, by the middle of the 2016.
- Python: Introduction of highly optimized Python support – specifically high performance SciPy and NumPy. Information about the current optimized Python distribution is at http://bit.ly/intel-python.
- Blueprint for your clusters: Intel’s HPC Scalable System Framework is a flexible blueprint for developing high performance, balanced, power-efficient and reliable systems. The performance and consistency this brings makes HPC easier to deploy, and that helps expand usage of HPC. I’m excited about the systems we will see come to market, in 2016, which utilize our HPC Scalable System Framework to help meet the needs of all types of HPC users, new and old.