CUDA.jl 4.0: Using CUDA-Accelerated Binaries and Libraries in JuliaCUDA.jl 4.0 was a breaking release that overhauled how users work with the CUDA toolkit in order to support the use of external CUDA-accelerated libraries and applications.
Dr. Tim Besard, JuliaHub software engineer, will explain these breaking changes, how to adapt your code or application, and how to make use of the new features and integrate with an existing CUDA-accelerated C application/library.
Dr. Besard also covers other recent improvements and changes to CUDA.jl, such as improved support for sparse arrays. This will make it possible to package existing GPU software and use it with CUDA.jl.
Participants will learn how to package a CUDA-accelerated application, how to use it in Julia with CUDA.jl and how to handle the CUDA toolkit in relation to that dependency.
Developers who are interested in CUDA and GPU computing should also watch the GPU Programming in Julia Webinar.
