Intel Software Tools Training Workshop

Dec 1, 2020, 10:00 AM → Dec 2, 2020, 5:00 PM Europe/Berlin

virtual

On December 1st and 2nd 10:00-17:00, a training for Intel Tools and Software for HPC is available (for free). The courses are taught by experts from Intel and include demo as well as Q&A sessions. 

The event is organized by the competence network HPC.NRW and Paderborn Center for Parallel Computing  (PC²).

No NDA is required to attend the workshop. The event will be held in a virtual way via a videoconference service.

 

Please Note: to register for this event you don't need an account in this Indico-System but you can just click on "Apply for participation" below.

Tuesday, December 1

10:00 AM → 10:15 AM Welcome and Introductions

This 1 day virtual course will start with a short update on the latest publicly available Intel Hardware Architectures and introduce the Intel Software development tools starting the various components of Intel Parallel Studio XE suites and the upcoming oneAPI development framework supporting heterogenous Intel environments.

10:15 AM → 11:00 AM The Intel Architecture for Software Developers

This session will offer insights into Intel hardware platforms tailored to the needs of software developers, software architects and HPC experts. Learn how Intel Software Development Tools will help you to achieve optimal performance via vectorization, memory access tuning, and threading.

11:00 AM → 12:00 PM Intel oneAPI – the Future of Intel Software Development

Modern workloads are incredibly diverse—and so are architectures. No single architecture is best for every workload. Maximizing performance takes a mix of scalar, vector, matrix, and spatial (SVMS) architectures deployed in CPU, GPU, FPGA, and other future accelerators. Intel oneAPI products will deliver the tools you need to deploy your applications and solutions across SVMS architectures.

12:00 PM → 1:00 PM Lunch Break

1:00 PM → 2:30 PM Intel Compilers – an Overview - with demos

C++ and Fortran compilers from Intel produce optimized code that takes advantage of the ever-increasing core count and vector register width in Intel processors. You will learn most important compiler optimizations helping to boost the performance of your application. A short demo will complement the talk and show a possible speed-up you can gain by controlling compiler options only.

2:30 PM → 3:30 PM Intel MKL – The Math Kernel Library - with demos

Intel Math Kernel Library is a key element in many applications for scientific, engineering, and data applications and many others. The session will cover the newest features implemented into the latest version and some of the newest coming updates. We will also describe to you the most efficient ways how to use Intel MKL properly and how to get the best performance out of your MKL applications.

3:30 PM → 4:30 PM Q&A

Wednesday, December 2

10:00 AM → 10:15 AM Welcome to Day 2 virtual Intel Tools Workshop

This second day virtual course will focus on demonstrating Intel Software Analysis Tools. We will start with the tools for shared memory environments (Intel VTune Profiler and Intel Advisor) and we show the high-level lightweight Application Performance Snapshot tool. Later we turn our attention to shared memory apps using Intel MPI for building and the Intel Trace Analyzer and Collector for analyzing MPI based applications. As time allows new features of the oneAPI development framework are shown.

10:15 AM → 11:15 AM Intel MPI Library - with demos

Get the best performance out of your MPI application by leveraging the Intel MPI library. Besides a general introduction on how to use Intel MPI properly, approaches for tuning the Intel MPI library will be shown.

11:15 AM → 11:45 AM Application Performance Snapshot - Intel VTune Profiler - with demos

Intel Application Performance Snapshot (APS) is the first step in a systematic analysis using Intel tools. APS identifies the hotspots of an application and recommends the most appropriate Intel tool for detailed analysis. A small demo will explain the basic steps of APS usage and result interpretation

11:45 AM → 1:00 PM Lunch break

1:00 PM → 2:00 PM Intel Advisor - Optimize high-performing code on modern computer architectures- with demos

SIMD (Single Instruction Multiple Data) instructions are crucial for optimal performance on modern Intel CPUs. On x86 architecture these extensions are known as SSE, AVX, AVX-2 and AVX-512. In general developers will rely on the compiler vectorization capabilities. But the automatic vectorization often needs support from the developer to deal with dependencies, alignment, data type conversions and other topics we will cover here. Compiler optimization reports provide some analysis but don’t give much guidance. Intel Advisor XE delivers an easy and precise way of vectorization analysis plus an evaluation of maximal potential performance for each loop (Roofline Analysis).

2:00 PM → 3:00 PM Intel VTune Profiler - Locate performance bottlenecks fast - with demos

Without the right data, you’re guessing about how to improve software performance and are unlikely to make the most effective improvements. The VTune Profiler collects key profiling data and presents it with a powerful interface that simplifies its analysis and interpretation. This talk will provide a general introduction to VTune while diving deeper using a memory bandwidth analysis example

3:00 PM → 4:00 PM Intel Trace Analyzer and Collector - MPI profiling for cluster applications -with Demos

Intel Trace Analyzer and Collector provides a detailed analysis and visualization of MPI usage in your application. The graphical display will help to understand bottlenecks of the application. It can also help to understand an unknown MPI algorithm. Another mode of ITAC is called Message-Checker (mc).  Running your application with mc will produce error messages or warnings when MPI related issues are detected. Both modes of ITAC are easy to be used but configuring for trace file reduction and user events are more advanced topics.

4:00 PM → 5:00 PM Q&A