Collecting and understanding performance and efficiency characteristics of applications is key for High Performance Computing. During this one day course with hands-on you will be equipped with advanced information about performance and efficiency metrics and how to easily obtain them using selected Intel tools. We will pick the pearls of each tool and advise you how to extract information about your application quickly and how to explain the extracted information for your application and target architecture. Since such focus areas arise on a node level we will primarily focus on that but also address how to work on a system/cluster level. Intel tools covered will be Intel(R) VTune(TM) Amplifier XE (together with "perf"), Intel Software Development Emulator and Intel(R) Advisor.
Since advanced performance analysis is architecture related, we use the available systems on the Salomon cluster, which will be used by the participants for the hands-on exercises. Salomon comprises nodes with Intel(R) Xeon(R) processors v3 (Haswell) and first generation Intel(R) Xeon Phi(TM) coprocessor (KNC). For the hands-on part participants are encouraged to bring their own applications for analysis. Please refer to the Prerequisites section below for this.
Purpose of the course (benefits for the attendees)
At the end of this course participants will be able to extract key metrics from applications using the described tools and methodologies, and make decisions about further performance tuning.
About the tutor
Georg Zitzlsberger formerly worked for Intel Deutschland GmbH (Germany). He has been a Technical Consulting Engineer for Intel(R) Software Development tools for many years before he recently joined IT4Innovations. In his new role as researcher he offers consulting services for Intel Software Development tools and Intel architectures throughout the IT4Innovations network.
|Wednesday 14 June 2017|
|9:30-9:45||Introduction & Agenda|
|9:45-10:45||The architectures in a nutshell|
|What is important to know for performance analysis?|
|What are typical perfromance bottlenecks?|
|System considered is Salomon: Intel(R) Xeon(R) processors v3 (Haswell) and first generation Intel(R) Xeon Phi(TM) coprocessor (KNC)|
|11:00-11:45||Intel(R) VTune(TM) Amplifier XE:|
|Obtaining key metrics from your application|
|Black Belt Guide (understand the metrics)|
|11:45-12:30||Apply Intel(R) VTune(TM) Amplifier XE to your application - Hands-On|
|12:30-13:30||Time for lunch|
|13:30-14:15||Intel(R) Advisor and Intel Software Development Emulator|
|Intel Advisor (vector tool)|
|Roofline-Model (arithmetic intensity and FLOPS)|
|SIMD & memory analysis|
|Simulate different micro-architectures|
|14:15-15:00||Apply Intel(R) Advisor and Intel Software Development Emulator to your application - Hands-On|
|15:15-16:45||How to use the tools at System/Cluster Level?|
|16:45-17:30||Final Hands-On with Q&A|
Participants are welcomed and advised to bring their own applications for the best training experience. For participants with no own applications, the trainer will provide prepared examples for the hands-on parts.
For participants without own applications, the trainer will provide examples.
Capacity and Fees
- Documentation for IT4Innovations' computer systems is available at https://docs.it4i.cz/.
- See links below how to get to the campus of VŠB - Technical University Ostrava and to the IT4Innovations building.