Topic outline
General
This course covers performance engineering approaches on the compute node level. Even application developers who are fluent in OpenMP and MPI often lack a good grasp of how much performance could at best be achieved by their code. This is because parallelism takes us only half the way to good performance. Even worse, slow serial code tends to scale very well, hiding the fact that resources are wasted. This course conveys the required knowledge to develop 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. We introduce the basic architectural features and bottlenecks of modern processors and compute nodes. Pipelining, SIMD, superscalarity, caches, memory interfaces, ccNUMA, etc., are covered. A cornerstone of node-level performance analysis is the Roofline model, which is introduced in due detail and applied to various examples from computational science. We also show how simple software tools can be used to acquire knowledge about the system, run code in a reproducible way, and validate hypotheses about resource consumption. Finally, once the architectural requirements of a code are understood and correlated with performance measurements, the potential benefit of code changes can often be predicted, replacing hope-for-the-best optimizations by a scientific process.Time: January 20-22, 2020Place: TBAPreliminary agenda:
Monday, January 20 13:00-13:30 Welcome & Intro 13:30-14:30 Basic computer architecture 14:30-15:00 Tools I: topology & affinity 15:00-15:15 Coffee break 15:15-16:15 Hands-on: simple benchmarks 16:15-17:00 Microbenchmarking
Tuesday, January 21 09:00-10:30 Roofline model basics 10:30-10:45 Coffee break 10:45-11:30 Tools II: hardware performance counters 11:30-12:00 Hands-on: dense MVM 12:00-14:00 Lunch break 14:00-14:30 Hands-on cont'd 14:30-15:30 Programming for SIMD & ccNUMA 15:30-15:45 Coffee break 15:45-17:00 Hands-on
Wednesday, January 22 09:00-10:30 Roofline case studies I 10:30-10:45 Coffee break 10:45-12:00 Roofline case studies II 12:00-12:30 Basic skills for Performance Engineering 12:30-13:30 Lunch break 13:30-15:30 Hands-on: code analysis 15:30-15:45 Coffee break 15:45-17:00 Outlook: the ECM performance model Day 1
Day 2
Day 3
Additional Material
For publications of the RRZE HPC group, see https://hpc.fau.de/research/publications/
Our HPC book (second edition coming up in 2020): Hager & Wellein: Introduction to High Performance Computing for Scientists and Engineers
Important links:
Intel® 64 and IA-32 Architectures Optimization Reference Manual
LIKWID tool suite: https://github.com/RRZE-HPC/likwid
LIKWID documentation Wiki: http://tiny.cc/LIKWID
LIKWID quick reference sheet
Kerncraft automatic Roofline/ECM modeling tool: https://github.com/RRZE-HPC/kerncraft
Online layer condition calculator: https://rrze-hpc.github.io/layer-condition/#calculator