Node-Level Performance Engineering
Program
This course teaches performance engineering approaches on the compute node level. “Performance engineering” as we define it is more than employing tools to identify hotspots and bottlenecks. It is about developing a thorough understanding of the interactions between software and hardware. This process must start at the core, socket, and node level, where the code gets executed that does the actual computational work. Once the architectural requirements of a code are understood and correlated with performance measurements, the potential benefit of optimizations can often be predicted. We introduce a “holistic” node-level performance engineering strategy and apply it to different algorithms from computational science. Architectural details that are relevant for performance, such as pipelining, SIMD, superscalarity, memory hierarchies, etc., are covered in due detail.
Participants must have basic knowledge in programming with Fortran or C and basic knowledge of OpenMP.
Registration
Registration is now open! Please register via the Indico conference management system.
Fee
Bachelor / Master Students from Hessen & Rheinland-Pfalz | 40 EUR |
PhD students from Hessen & Rheinland-Pfalz | 60 EUR |
Members of German universities and public research institutes | 60 EUR |
Others | 580 EUR |
Fee includes coffee breaks, but not lunch & dinner.
Social Events
On the first evening, we plan a dinner (self-paying) at “Zum Storch am Dom”
Cluster Computing Course
The Cluster Computing Course is for participants of Node-Level Performance Engineering.
Friday, August 30 2019
Location
Goethe Universität, Campus Riedberg (Frankfurt am Main)
Program
Cluster facts of the GOETHE-HLR & FUCHS cluster:
-
Hardware resources
-
File system
-
Environments modules
-
Partitions on the cluster
-
Architecture of the partitions
Batch Usage:
SLURM is the job scheduler installed on GOETHE-HLR & FUCHS cluster. The session teaches attendees
-
how to prepare a submission script,
-
how to submit, monitor, and manage jobs on the clusters,
-
theory about resource and CPU management.
Travel Information and Accommodation
See our directions, the campus map, and the entrance and room 114 of building N100.
Public transportation
From main railway Station “Hauptbahnhof” with S-Bahn S1 - S9 to “Hauptwache”, then with U-Bahn U8 (direction Riedberg) to “Uni Campus Riedberg”.
Hotel Recommendation
Contact
Anja Gerbes, +49 (0)69 798-47356
This course is organized by HKHLR and CSC, Goethe University Frankfurt in cooperation with RRZE &
Agenda
Day 1:
- 9:30 Welcome - Intro
- 9:45 Computer architecture for software developers
- 10:45 Coffee Break. 15m
- 11:30 Performance Engineering Basics
- 12:00-13:00 Lunch Break
- 13:00 Tools for Performance Engineering 1
- 13:30 Exercise 1: The Bandwidth Benchmark
- 15:15 Coffee Break 15m
- 15:30 Roofline Model: Basics
- 17:00 End of day
Day 2:
- 9:00 Tools for Performance Engineering 2
- 9:45 Optimal use of parallel resources: SIMD, ccNUMA
- 10:30 Coffee Break 15m
- 11:15 Exercise 2: Dense Matrix Vector Multiplication
- 12:00-13:00 Lunch Break
- 13:00 Performance Engineering Basic Skills
- 14:00 Case Study: Roofline Model Jacobi smoother
- 15:15 Coffee Break 15m
- 15:30 Execise 3: Analysis of miniMD proxy app
- 17:00 End of day