In this workshop, we will demonstrate three major packages for programming GPUs in Julia (CUDA.jl, AMDGPU.jl, oneAPI.jl), and the different programming models, tools and APIs that these packages support.

Enzyme (https://enzyme.mit.edu) is a reverse mode auto-differentiation tool that performs automatic differentiation over LLVM intermediate representation and synthesis high-performance reverse-mode functions. We will discuss how Enzyme.jl integrates with the Julia compiler and special considerations required for differentiating a dynamic programming language such as Julia.

This talk will present scaling and performance of the Oceananigans.jl ocean model on CPU and GPU systems. Oceananigans.jl is an all Julia code that is designed to study geophysical fluids problems ranging from idealized turbulence to planetary scale circulation. It uses the KernelAbstractions.jl package to support CPU and GPU single address space parallelism. It uses MPI.jl, to support multi-node and multi-GPU parallelism. MPI.jl is used both directly and through PencilArrays.jl.

The JuliaHPC community as a group maintains the infrastructure for using Julia in high-performance computing. In this BoF we invite newcomers, application developers, and maintainers to join us for an informal discussion around the state of Julia in HPC.

This is a BoF to talk about the various GPU programming packages in Julia:

• CUDA.jl
• AMDGPU.jl
• oneAPI.jl
• KernelAbstractions.jl
• GPUArrays.jl
• GPUCompiler.jl
• ...

If you have any thoughts or questions about these packages, or other approaches to GPU programming in Julia, please join this BoF to chat about it!