Categoria: Seminari e Convegni
Stato: Archiviata
13-14 November 2019

Artificial Intelligence & High-Performance Computing: Performance Optimization with Intel Software Tools

Disma - Buzano Room

In this 2-day Hands-on workshop, a unique opportunity will be provided to learn techniques, methods and solutions on how to improve code, how to enable new features of the Intel processors and to use new tools like the Roofline model to visualize the potential benefits of an optimization process for HPC and AI applications.

On the first day of this workshop the Intel SW development tools and the new, upcoming cross platform development concept oneAPI and the code optimization process will be introduced.
The attendees will be supported with hands-on exercises using a many-body kernel and learn how to enable vectorization with simple pragmas and more effective techniques, like changing data layout and memory alignment. The work will be then guided by the hints from the Intel Compiler reports and using the Intel Advisor. Furthermore, the participants will get insights into Intel VTune Amplifier and of Intel Application Performance Snapshot as tools for investigating and improving the performance of HPC application.

On the second day the latest micro-processor hardware architectures and how the developers can efficiently use modern HPC hardware will be described, in particular the Intel Deep Learning Boost with AVX-512VNNI instructions.
The attention will be then focused on data analytics techniques, such as Machine Learning and Deep Learning, which become the key for gaining insight into the incredible amount of data generated by scientific investigations (simulations and observations). An overview on the most known machine learning algorithms for supervised and unsupervised learning will follow. With small example codes it will be showed how to implement such algorithms using the Intel Distribution for Python, and which performance benefit can be obtained with minimal effort from the developer perspective.

Simple algorithms will be implemented, like K-Means and PCA, using the Intel Data Analytics Acceleration Library (Intel-DAAL) and showed how to scale the workload on HPC systems using the Intel MPI library. The seminar will also cover how to accelerate the training of deep neural networks with Tensorflow, thanks to the highly optimized Intel Math Kernel Library (Intel MKL). Techniques on how to leverage deep neural network training on multiple nodes on distributed x86 HPC systems not requiring GPUs will be demonstrated.

For further information please visit the website
Registration is mandatory at