Current maintainer	In relation to lecture by
youwuyou (youwuyou@ethz.ch)	CSC Finnland

CSC Summer School in High-Performance Computing 2022

This is the material repository for the high-performance computing summer school by CSC - Finnish IT Center for Science.

Once forked to yourself, you can sync with the original repository (in case of updates) by running:

git pull https://github.com/csc-training/summerschool.git

Exercise Overview

Topic	Name	Finished
Introduction	Hello world	✔️
Point-to-point communication	simple message exchange	✔️
Point-to-point communication	simple parallel pi	✔️
Communication patterns	message chain	✔️
Communication patterns	parallel pi	✅
Communication patterns	broadcast and scatter
Collective operations	collective operations	✔️
Custom communicators	communicators and collectives
Heat equation	parallel heat equation solver
Heat equation	heat equation solver with sendrecv
Heat equation	collective communication in heat equation solver
Simple derived types	simple datatypes	✅
advanced derived types	communicating struct/derived type	✔️
advanced derived types	modifying extent
process topologies	cartesian grid process topology	✔️
process topologies	message chain with Cartesian communicator
Non-blocking communication	Non-blocking message chain
Non-blocking communication	message chain with persistent communication
Heat equation	heat equation with cartesian communicator
Heat equation	non-blocking communication in heat equation
Heat equation	2D-decomposed heat equation

Topic	Name	Finished
Intermediate level	Parallel I/O with Posix	✅
Intermediate level	Simple MPI-IO
High-level	HDF5 example
	Bonus: Checkpoint + restart with MPI-IO

Topic	Name	Finished
Intro OpenMP	hello world	✔️
Intro OpenMP	work sharing and vector addition	✔️
Library functions	OpenMP library functions	✔️
Data sharing	Data sharing and parallel regions	✔️
Data sharing	race condition in parallel sum	✔️
Reductions and execution controls	reduction	✔️
Reductions and execution controls	execution controls	✔️
Heat equation	fine grain parallelization
Heat equation	coarse grain parallelization
Hybrid MPI+OpenMP programming	hybrid hello world
Hybrid MPI+OpenMP programming	multiple thread communication
OpenMP tasks	simple tasking
OpenMP tasks	Parallelizing Mandelbrot with tasks
OpenMP tasks	Parallel Fibonacci
Hybrid heat equation	hybrid heat equation

Topic	Name	Finished
Intro to GPUs in HPC	device query	✅
Basics of OpenMP offloading	Hello world	✔️
Basics of OpenMP offloading	vector addition
Basics of OpenMP offloading	simple offloading for Laplace equation
Controlling data movement	dot product
Controlling data movement	sum and dot product
Controlling data movement	offloading the heat equation
Unstructed data	unstructed data in heat equation
Device functions	device functions
Interoperability with libraries	using CUDA library
Asynchronous operations	Asynchronous operations
Using multiple GPUS	multi-GPU parallelization for heat equation

Topic	Name	Finished
Basics of HIP programming	hello world	✔️
Basics of HIP programming	Kernel: saxpy	✔️
Basics of HIP programming	Kernel: copy2d	✔️
Synchronisation and streams	investigating streams and events	✔️
Memory management	memory management strategies	✅
Memory management	unified memory and structs	✔️
Fortran and HIP	Hipfort: saxpy	✔️
Multi-GPU programming	vector sum on two GPUs without MPI	✔️
Multi-GPU programming	ping-pong with multiple GPUs and MPI
Multi-GPU programming	peer to peer device access
Multi-GPU programming	Bonus: Heat equation with HIP

for more theoretical aspects see my side-notes