Monday, November 27, 2017, 10:00am – 4pm  (the event will be held in German language)

Das HPC-Admintreffen dient der Förderung des Austauschs unter den HPC-Administrierenden in NRW.

Probleme und Lösungsansätze im Bereich HPC-Administration werden diskutiert.

Folgende Themen sind für die Agenda geplant:

• Singularity-Container im HPC-Cluster
• Wechsel von Workloadmanagement Systemen
• Lösungen für die bundesweite Öffnung/Authentifizierung
• Austausch zu NFS4/Kerberos, parallelen Filesystemen, Backup-Konzepten, SW-Stack Anwendungssoftware.

Anmeldung

Die Anmeldung ist geschlossen

Kontakt
hpc-admintreffen-info@lists.rwth-aachen.de

Introduction

The NEC SX-Aurora TSUBASA is a new vector processor based accelerator for HPC workloads. Since this vector engine (VE) provides a very high memory bandwidth (1.2 TB/s), it is a very promising architecture for memory bound applications. The VE is provided as a PCIe card in a x86-64 server. Each VE is equipped with 8 cores. This workshop provides detailed information about the Aurora Architecture. Experts from NEC will present the programming environment and the different  execution models including hints for performance tuning. The attendees of the workshop have the opportunity to explore the system by working on provided hands-on or bringing their own code.

Details

Date: Thursday 2nd – Friday 3rd May 2019.

Location: IT Center, RWTH Aachen University, Germany

Room: Seminar Room 3, Kopernikusstraße 6

Organizer: NEC and RWTH Aachen University

Agenda

May 2nd Thursday

                11:30 – 13:00                        Registration and Lunch

                13:00 – 13:45                        Aurora Architecture Deep Dive (Shintaro Momose)

                13:45 – 14:30                        VEOS basics, tweaking and news (Erich Focht)

                                                             Part I: VEOS high level design

                                                             Part II: VEOS Intro

                14:30 – 14:45                        Break

                14:45 – 15:45                        User contributions: 10-15min each

                                                             Aurora Experiments (Kurt Lust)

                                                             Aurora: User Experience (Daniel Schürhoff)

                                                             Using the NEC Aurora Workshop System for APES framework (Kannan Masilamani)

                                                             VASP and NEC SX (Bernhard Eck)

                                                             Demands on HPC/Aurora in CFD (Arne Rempke)

                15:45 – 16:15                       Tutorial part 1 (Christian Weiss)

                16:30 – 18:00                        Hands-on 1 (Christian Weiss, Felix Uhl, Rene Puttin)

                —

                19:00 –                                  Dinner sponsored by NEC                                                                      

May 3rd Friday

                08:30 – 09:00                       Tutorial part 2 (Christian Weiss)

                09:00 – 10:30                       Hands-on 2 (Christian Weiss, Felix Uhl, Rene Puttin)

                10:30 – 10:45                       Break

                10:45 – 11:30                       Aurora offloading (Erich Focht)

                11:30 – 12:00                       AI on Aurora (Erich Focht)

                12:00 – 13:00                       Q & A & Finger Food (all) 

Bring Your Own Code

All attendees have the opportunity to bring their own code and get assistance from the NEC experts for code porting and performance tuning. If you want to bring your own code please keep in mind to prepare a small, but representative data set. We have 8 VE (8 cores and 48 GB memory each) available. Since we have to share the VEs with all attendees the data set should not use more than 8-16 MPI processes and less than 48 to 96 GB. Furthermore, the runtime should be about 10 minutes (or less). If you want to prepare your code or data in our environment before the workshop begins, please contact Tim Cramer <cramer@itc.rwth-aachen.de> for an individual test account.

Dinner

The Dinner will be in the restaurant Palladion in the historical city of Aachen. Address:

Restaurant Palladion
Schmiedstraße 3
52062 Aachen

Registration

Please register here >>> Please register until Friday, April 12, 2019 April 19, 2019

Contact

Tim Cramer

hpcevent@itc.rwth-aachen.de or

Oliver Tennert

oliver.tennert@emea.nec.com

Travel Information

Please make your own hotel reservation. You may find a list of hotels in Aachen on the web pages of Aachen Tourist Service.  We recommend that you try to book a room at the INNSIDE Aachen, Novotel Aachen City, Mercure am Graben or Aachen Best Western Regence hotels. These are nice hotels with reasonable prices within walking distance (20-30 minutes, see city map) from the IT Center through the old city of Aachen. An alternative is the IBIS Aachen Marschiertor hotel, located close to the main station, which is convenient if you are traveling by train and also prefer to commute to the IT Center by train (4 trains per hour, 2 stops).

Workshop on Artificial Intelligence    Thursday, December 06, 2018  10:00h to 16:30h IT Center, RWTH Aachen University, Kopernikusstrasse 6 / Seminarraum 003  sponsored by: The workshop was fully booked !
After  the workshop, please provide your feedback here >>>   (click on ” Respond to this Survey“)

Introduction

Artificial intelligence (AI) technology like Machine learning (ML) and Deep learning (DL) are today more and more used to solve a wide range of use cases.

The HPC cluster at RWTH can be used for applications using various AI methods.

This workshop gave the participants an overview on AI covering different methods like classical ML and DL, and introduced the use of DL on (Intel) CPU based HPC cluster. The presentations were rounded-up by live demos.

Mr Stephan Gillich (Dir. AI and Technical Computing GTM EMEA, Intel Deutschland GmbH) and his colleagues Atanas Atanasov and Sarosh Quraishi (Technical Solution Specialists, Intel Deutschland GmbH) were complemented by experts from the HPC Team  of the IT Center of RWTH Aachen University.

Agenda

• 10:00 – 10:45  Welcome,  RWTH goals and specific interests (RWTH) – slides1, slides2
  
• 10:45 – 12:00  AI overview and deep learning frameworks on a cluster (Intel) – slides
  
  • • • We will cover a few fundamental things on AI and the various techniques like ML and DL and take a closer look at how we can help customers on their AI journey
• 12:00 – 13:30:  lunch break
• 13-30 – ~15:30:  Faster ML and DL with optimised frameworks (Intel)
  • • • After first introducing libraries and tool for classical Machine Learning and Analytics and Deep Learning, we will mainly cover and demonstrate technical best practices for faster time to solution, including scale-out training leveraging the performance of multiple datacenter nodes – slides
        
      • demo/hands-on – slides
        
• 15:30 – 15:45  Coffee Break
• 15:45 – 16:30  Inference solutions. (Intel)
  
  • • • The Inference side of deep learning using an optimized framework setup including  the OpenVINO™ (Open Visual Inference & Neural network Optimization)  – slides
        
• 16:30 – 17:00  Q&A (all)
  

Prerequisites and your Preparation

Participants from RWTH Aachen University were welcome, but also participants from other universities or industrial companies.

The presentations were given in English.

Course Material

Welcome – C. Terboven (pdf, 0.7 MB)

Introduction – M. Widmer (pdf , 4MB)

Slides of the morning session (pdf, 6 MB)

Optimized-Frameworks (pdf, 4.7 MB)

Slides for the afternoon hands-on session (pdf, 1.5 MB)

OpenVINO-Deep-Dive (pdf, 1,7 MB)

Lab (pdf, 3 MB)

Lab Sources (GigaMove Download Link to zip file (11 MB)  – until Jan 24, 2019 ) expired!!