Directive Based GPU Programming Models

Overview of Modular Lesson Material

This lesson module targets developers who want to run computations on GPUs without writing intricate, low-level kernel code like CUDA or OpenCL. Instead, the lesson teaches high-level GPU abstraction by utilizing compiler pragmas to automatically parallelize existing sequential code and offload data. This design choice allows developers to focus on writing clean code in familiar languages like C/C++ or Fortran while maintaining high computational performance.

The lesson covers everything from foundational abstractions to complex multi-accelerator scaling patterns. Core episodes focus on established industry-standard frameworks, primarily evaluating OpenACC and OpenMP offloading capabilities to bridge ease of development with hardware efficiency. Beyond single-GPU optimization, the syllabus advances into multi-GPU environments, mapping out the integration of directive-based models with MPI (Message Passing Interface) for cluster-wide communications.

Ultimately, this educational effort addresses an ongoing challenge in modern high-performance computing (HPC): migrating massive codebases to modern supercomputers. Because the manual translation of legacy sequential algorithms into hardware-specific kernels requires deep architecture knowledge, directive-based abstraction serves as an indispensable tool for onboarding scientists.

Prerequisites

  • Familiarity with one or more programming languages like C/C++ or Fortran is recommended

  • Basic understanding of parallel computing concepts

Table of Contents

Setup

Episodes

Reference

Learning outcomes

This material is for all researchers and engineers who work with large or small datasets and who want to learn powerful tools and best practices for writing more performant, parallelised, robust and reproducible data analysis pipelines.

By the end of this module, learners should:

  • Incrementally transform sequential programs into GPU-accelerated versions using directives

  • Understanding how directives translate to GPU execution

Credit

Don’t forget to check out additional course materials from XXX. Please contact us if you want to reuse these course materials in your teaching. You can also join the XXX channel to share your experience and get more help from the community.

License

CC BY-SA for media and pedagogical material

Copyright © 2025 XXX. This material is released by XXX under the Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0).

Canonical URL: https://creativecommons.org/licenses/by-sa/4.0/

See the legal code

You are free to

  1. Share — copy and redistribute the material in any medium or format for any purpose, even commercially.

  2. Adapt — remix, transform, and build upon the material for any purpose, even commercially.

  3. The licensor cannot revoke these freedoms as long as you follow the license terms.

Under the following terms

  1. Attribution — You must give appropriate credit , provide a link to the license, and indicate if changes were made . You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.

  2. ShareAlike — If you remix, transform, or build upon the material, you must distribute your contributions under the same license as the original.

  3. No additional restrictions — You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.

Notices

You do not have to comply with the license for elements of the material in the public domain or where your use is permitted by an applicable exception or limitation .

No warranties are given. The license may not give you all of the permissions necessary for your intended use. For example, other rights such as publicity, privacy, or moral rights may limit how you use the material.

This deed highlights only some of the key features and terms of the actual license. It is not a license and has no legal value. You should carefully review all of the terms and conditions of the actual license before using the licensed material.

MIT for source code and code snippets

MIT License

Copyright (c) 2025, EVITA project, Cristian-Vasile Achim, Hicham Agueny, Andrey Alekseenko, Richard Darst, Karim Elgammal, Francesco Fiusco, Juan de Gracia, Johan Hellsvik, Jaro Hokkanen, Erik Holmström, Qiang Li, Wei Li, Daniel Medeiros, Ashwin Vishnu Mohanan, Tapish Narwal, Pedro Ojeda May, Yann Pfau-Kempf, Stephan Smuts, Stepas Toliautas, Apostolos Vasileiadis, Yonglei Wang, and Kjartan Thor Wikfeldt

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Note

To module authors: For code you may use any OSI-approved license as mentioned in https://spdx.org/licenses/, such as Apache License 2.0, GNU GPLv3, MIT. Please make sure to update the deed above and LICENSE.code file accordingly.