2025-07-25 –, Main Room 5
In this talk we'll summarize and demonstrate some of the improvements made to the CUDA.jl package over the past year; including new features in the compiler, memory management, and device programming stack; as well as updates about the support for various CUDA libraries. Practical examples will be provided to show the benefits of this work for both end users and developers of packages which rely on CUDA.jl.
CUDA.jl is one of the most mature and widely used libraries in the JuliaGPU family. Much of the HPC, SciML, and graphics stack depends on it. In this talk we will communicate many of the improvements the development team has made over the past year and demonstrate why and how they're useful for a variety of audiences. We'll cover both the "core" CUDA stack for integrating with the device, its driver, and its runtime, as well as the improvements to the wrappers for various application libraries such as CUBLAS, CUFFT, and more.
I am a Julia contributor since 2015. I work mostly on GPUs, quantum packages, and linear algebra.