Welcome and introduction

Self presentations One minute-one slide self presentations

Opening Session

• Alberto Pace (CERN)

Preparing for the HL-LHC computational challenge

• Danilo Piparo (CERN)

In this talk we will introduce some basic concepts related to HEP data processing and analysis workflows, seeing them in action in the context of LHC experiments. We’ll also talk about the evolution of the LHC accelerator and experiments. We’ll characterise at a high level what are the consequences of those upgrades for the HEP data processing software, in particular in the context of an evolving hardware and computing infrastructure.

Announcements Room: Room m25/26

CPU Hardware Architecture and Evolution

• Guilherme Amadio (CERN)

CPU Architecture and High Performance - lecture 1 - Hardware evolution of the CPU - Memory hierarchy, caching, NUMA - Microarchitecture of modern CPUs

Performance Analysis on Modern CPUs

• Guilherme Amadio (CERN)

CPU Architecture and High Performance - lecture 2 - Performance analysis tools for Linux - CPU features for performance analysis - Top-down microarchitecture analysis

CPU Architecture and High Performance - exercise 1

• Guilherme Amadio (CERN)

Writing parallel software

• Sebastien Ponce (CERN)

Room: Room m25/26 Parallel and Optimised Scientific Software - lecture 1 - Amdahl's and Gustafson's laws - Asynchronous execution - Finding concurrency, task vs. data parallelism - Using threading in C++ and Python, comparison with multi-process - Resource protection and thread safety - Locks, thread local storage, atomic operations

Writing efficient software

• Sebastien Ponce (CERN)

Room: Room m25/26 Parallel and Optimised Scientific Software - lecture 2 - Why Python and C++ ? - Recent evolutions: C++ 11/14/17 - Modern features of C++ related to performance - Templating versus inheritance, pros and cons of virtual inheritance - Python 3, and switching from Python 2

Announcements Room: Room m25/26

Low-level Performance Optimization Guidelines

• Guilherme Amadio (CERN)

Room: Room m25/26 CPU Architecture and High Performance - lecture 3 - Main sources of performance bottlenecks - Floating point arithmetics performance - Advanced low-level performance tuning

CPU Architecture and High Performance - exercise 2

• Guilherme Amadio (CERN)

Room: Room m25/26

Optimizing existing large codebase

• Sebastien Ponce (CERN)

Room: Room m25/26 Parallel and Optimised Scientific Software - lecture 3 - Measuring performance, tools and key indicators - Improving memory handling - The nightmare of thread safety - Code modernization and low level optimizations - Data structures for efficient computation in modern C++

Data-Oriented Design

• Guilherme Amadio (CERN)

Room: Room m25/26 CPU Architecture and High Performance - lecture 4 - Principles of data-oriented design - Memory access and data-type profiling - Data structure performance optimization

Managing Large Databases: Challenges and Lessons

• Andrzej Nowicki (CERN)

Room: Room m25/26 With around 100 Oracle databases and several PB of data, CERN faces numerous intriguing challenges and unique caveats in regards to managing the database infrastructure. This lecture aims to provide an overview of both past and current issues and obstacles encountered in the realm of database systems at CERN. By doing so, it seeks to offer valuable insights into potential problems to avoid and effective solutions to address encountered difficulties.

Scientific computing on heterogeneous architectures

• Daniel Hugo Campora Perez (NVIDIA)

Room: Room m25/26 Programming for Heterogeneous Architectures - lecture 1 - Introduction to heterogeneous architectures and the performance challenge - From general to specialized: Hardware accelerators and applications - Type of workloads ideal for different accelerators - Trade-offs between multi-core and many-core architectures - Implications of heterogeneous hardware on the design and architecture of scientific software - Embarrassingly parallel scientific applications in HPC and CERN

Optimizing existing large codebase - exercise 1

• Sebastien Ponce (CERN)

Room: Room m25/26 Optimisation of an existing, production grade large codebase

Announcements Room: Room m25/26

Optimizing existing large codebase - exercise 2

• Sebastien Ponce (CERN)

Room: Room m25/26 Optimisation of an existing, production grade large codebase

Transport to Nikola Tesla museum - Departure at 14.15 sharp! Room: Room m25/26

Free time in Belgrade - visits, shopping, resting in a cafe... Room: Room m25/26 The Tesla museum is located in Vracar, one of the nicest neighborhoods in Belgrade, the Krunska street where the Museum is located is one of the most famous streets in Belgrade. Suggested activities, free to chose from - A must see: Saint Sava Temple, the 4th biggest Eastern Orthodox church in the world, only about a 10 min walk from the Museum. - Tasmajdan Park, with St. Mark’s Church, 10 min from museum. - Walk to Kalemegdan park, visit to the Belgrade Fortress (2,3 km from museum) https://www.beogradskatvrdjava.co.rs/?lang=en - Shopping around in old Belgrade - Go to a cafe or pub, a good cafe chain is Kafeterija Meeting at 19.30 at Restaurant Smokvica in Molerova 33 (6 min by foot from Nikola Tesla museum)

Performant programming for GPUs

• Daniel Campora (NVIDIA)

Room: Room m25/26 Programming for Heterogeneous Architectures - lecture 2 - From SIMD to SPMD, a programming model transition - Thread and memory organization - Basic building blocks of a GPU program - Control flow, synchronization, atomics

Performant programming for GPUs

• Daniel Campora (NVIDIA)

Room: Room m25/26 Programming for Heterogeneous Architectures - lecture 3 - Data locality, coalesced memory accesses, tiled data processing - GPU streams, pipelined memory transfers - Under the hood: branchless, warps, masked execution - Debugging and profiling a GPU application

Announcements

Practical vectorization

• Sebastien Ponce (CERN)

Room: Room m25/26 Parallel and Optimised Scientific Software - lecture 4 - Measuring vectorization level - What to expect from vectorization - Preparing code for vectorization - Vectorizing techniques in C++: intrinsics, libraries, autovectorization

Practical vectorization - exercise

• Sebastien Ponce (CERN)

Room: Room m25/26

Programming for heterogeneous architectures - exercise

• Daniel Campora (NVIDIA)

Room: Room m25/26

Programming for heterogeneous architectures - exercise

• Daniel Campora (NVIDIA)

Room: Room m25/26

Debrief of exercises Room: Room m25/26

Design patterns and best practices

• Daniel Campora (NVIDIA)

Room: Room m25/26 Programming for Heterogeneous Architectures - lecture 4 - Good practices: single precision, floating point rounding, avoid register spilling, prefer single source - Other standards: SYCL, HIP, OpenCL - Middleware libraries and cross-architecture compatibility - Reusable parallel design patterns with real-life applications

Announcements

Programming for heterogeneous architectures - exercise

• Daniel Campora (NVIDIA)

Room: Room m25/26

Programming for heterogeneous architectures - exercise

• Daniel Campora (NVIDIA)

Room: Room m25/26

Exam Room: Room m25/26

Closing Session

CERN outreach

• Sebastien Ponce (CERN)
• Alberto Pace (CERN)

Departures Room: Room m25/26