List-of-topics.md

Introduction to GPU programming model

• execution model
• hardware to software mapping (CU --> block, ...)
• memory hierarchy
• streams

cpp concepts

• classes, structures
• templates
• lambdas
• functors
• smart pointers
• namespace
• C++ stdpar

SYCL ( https://enccs.github.io/sycl-workshop/ )

• queues:

  • device association
  • device info
  • behaivior

• expressing parallelism:

  • lambdas
  • range
  • nd_range

• memory:

  • buffers
  • USM
  • local share memory

• dependencies:

• internal profiling and debugging:

• cuda to sycl: CUDA heat equation as an exercise

• installation on Mahti and LUMI (hipsycl/AdaptiveCpp & (intel llvm))

SYCL and outside libraries

• SYCL & MPI
• SYCL & CUDA/HIP libraries

Kokkos Core

• Short intro to inline building and programming

Possible exercises:

• selecting device in a node with several GPUs
• SYCL + mpi for computing pi (https://www.intel.com/content/www/us/en/developer/articles/technical/compile-and-run-mpi-programs-using-dpcpp-language.html)
• syclomatic for the heat equation
• calling other libraries (not necessarily portable)
• check examples from here https://github.com/oneapi-src/oneAPI-samples/tree/master/DirectProgramming/C%2B%2BSYCL

Sources

• https://sycl.tech/
• https://link.springer.com/book/10.1007/978-1-4842-9691-2
• https://tinyurl.com/book-SYCL
• https://github.com/Apress/data-parallel-CPP/tree/main
• git clone --recursive --branch sc23 https://github.com/codeplaysoftware/syclacademy.git
• https://www.youtube.com/watch?v=IeOnlBXTdn4&list=PL46sP9LM8GsyHAxj1k7MbWrv5f5SlMpIF
• https://enccs.github.io/sycl-workshop/
• https://enccs.github.io/gpu-programming/
• https://www.dropbox.com/scl/fo/bmgeetfr31i5vcz7j5sjw/h?rlkey=qg0tdg14tx65n0kxpxbz50bzo&dl=0
• https://www.intel.com/content/www/us/en/developer/tools/oneapi/training/migrate-from-cuda-to-cpp-with-sycl.html#gs.0zbdmk
• https://documentation.sigma2.no/code_development/guides_gpu.html#dev-guides-gpu
• https://fs.hlrs.de/projects/par/events/2023/intel-oneapi/
• https://fs.hlrs.de/projects/par/events/2023/GPU-AMD/ (some Kokkos)
• https://github.com/oneapi-src/oneAPI-samples/tree/master/DirectProgramming/C%2B%2BSYCL/Jupyter/oneapi-essentials-training
• https://github.com/oneapi-src/oneAPI-samples/tree/master/DirectProgramming/C%2B%2BSYCL/Jupyter/gpu-optimization-sycl-training
• https://github.com/oneapi-src/oneAPI-samples/tree/master/DirectProgramming/C%2B%2BSYCL/Jupyter/sycl-performance-portability-training
• https://github.com/oneapi-src/oneAPI-samples/tree/master/DirectProgramming/C%2B%2BSYCL/Jupyter/cuda-to-sycl-migration-training
• heat equation in cuda: https://github.com/cschpc/heat-equation/tree/main/cuda
• heat equation in sycl: https://github.com/ENCCS/gpu-programming/tree/main/content/examples/stencil/sycl
• heat equation at intel: https://www.intel.com/content/www/us/en/developer/articles/technical/solve-a-2d-heat-equation-using-data-parallel-c.html
• cuda training repository at CSC: https://github.com/csc-training/CUDA
• https://www.intel.com/content/www/us/en/developer/articles/technical/solve-a-2d-heat-equation-using-data-parallel-c.html
• https://doku.lrz.de/files/17826165/16942343/2/1686045904023/syclomatic-lrz-workshop+1.pdf
• hipstdpsar: https://gpuopen.com/learn/amd-lab-notes/amd-lab-notes-hipstdpar-readme/