Free

Intel® Application Optimization and Vectorization hands-on trainings

Event Information

Share this event

Date and Time

Location

Location

Cologne Chocolate Museum

Am Schokoladenmuseum 1A

50678 Köln

Germany

View Map

Friends Who Are Going
Event description

Description

You are kindly invited to the following 2 Intel Software Development Tools hands-on training events, which can be booked individually:

- A specialized 1-day software optimization training (24.4.) – making use of Intel VTune Amplifier (Performance Analyser)

- And/or a 1-day vectorization training (25.4.) – using the Intel Advisor and the Intel C/C++ Compilers

lead by development tools experts from the Intel Software Developers Product Division.


You can register either for the 1st day, the 2nd day or both days.


Application Optimization training:
The training targets Software developers using C/C++/Fortran/Python and related analysis tools and focuses on industry oriented hands-on labs using Intel® XeonTM servers. Lab training will be conducted on Linux. Notebooks will be provided by Intel in order to analyse data collected on the Intel® XeonTM Servers.

The participants will learn how to achieve significant performance improvements and create more reliable serial and parallel applications.

Agenda for the Application Optimization training (24.4.):

09:30 - 10.00: Registration & Coffee

10:00 - 11:00: Introduction to Intel® VTune™ Amplifier XE
- Demo Lab 1: Basic Hotspot analysis - Finding Performance Hotspots

11.00 - 11:15: Coffee Break

11:15 - 12:30:
- Demo Lab 2: Concurrency Analysis
- Demo Lab 3: Locks and Waits Analysis - Identifying Parallelism issues
- Demo Lab 4: Instrumenting code - User APIs and Tasks

12:30 - 13:30: Lunch Break

13:30 - 15:30: Intel Core and Intel Xeon microarchitectures overview
-
Introduction to Performance Monitoring Unit
- Event Based Sampling (EBS) Technology

- Event Based Performance Analysis with VTune Amplifier
Performance Analysis Methodology
- Identifying hotspots, Determine efficiency of hotspots and Fixing issues


Advanced hands-on activities:
- Advanced lab 1 – Find the performance hotspot using Advanced Hotspot Analysis
- Advanced lab 2 – Determining code efficiency, find stalls using General exploration

15:30 - 15:45: Coffee Break

15:45 - 17:45:
-Advanced Lab 3 - Finding memory issues
- Advanced Lab 4 - Find compuational issues
- Advanced Lab 5 - Analyzing brancing issues

Lab (new): Tuning Python Code with Intel® VTune™ Amplifier XE
Overview of vectorization principles

17:45 - 18:00: Q & A session


Vectorization training:
You will learn how to optimize SIMD codes with Intel Vectorization Advisor. We will cover these areas:
- What is vectorization and how can I vectorize my program?
- How can I understand if my program is vectorized (Advisor tutorial code)?
- What affects vectorization effectiveness (Compiler vectorization tutorial code)?
- How to improve vectorization effectiveness with Intel Advisor?
- Vectorizing HPC applications. Roofline model.

Agenda for the Vectorization training (25.4.):

09:30 - 10.00: Registration & Coffee

10:00 - 11:00: What is vectorization? How can I vectorize my program?
Parallel program structure. Amdahl law.
SIMD and vector registers. SSE, AVX, AVX512
Benchmarking
Vectorization approaches:
- Auto vectorisation with compiler
- SVML, MKL, IPP, DAAL libraries for program vectorization
- Vectorisation with OpenMP 4.0-5.0
- Use of intrinsics and assembly for vectorization

11:00 - 11:15: Coffee break

11:15 - 12:00: How can I understand if my program is vectorized? (Advisor tutorial code)
Understanding Vectorization Advisor grid
Peeling, Body, Reminder
Multi-versioning
Understanding compiler optimization report

12:00 - 13:00: Lunch Break

13:00 - 14:30: What affects vectorization effectiveness? (Compiler vectorization tutorial code)
Loops structure for vectorization
Data dependencies
Data access pattern
Data alignment
Vectorization of conditional statements and masking

14:30 - 14:45: Coffee Break

14:45 - 16:15: Vectorization using Intel compiler (Small code snippets)
Compiler option Pragma vector, ivdep
OpenMP 4.* and pragma SIMD
Vectorization of Fortran applications

16:15 - 17:00: How to improve vectorization effectiveness with Intel Advisor? (Advisor tests)

17:00 - 17:45: Vectorizing HPC applications and Roofline model (MKL vectorization, Roofline tutorial)

17:45 - 18:00: Q & A session

Share with friends

Date and Time

Location

Cologne Chocolate Museum

Am Schokoladenmuseum 1A

50678 Köln

Germany

View Map

Save This Event

Event Saved