“You can now write your CUDA kernels in Julia, albeit with some restrictions, making it possible to use Julia’s high-level language features to write high-performance GPU code.”
“The programming support we’re demonstrating here today consists of the low-level building blocks, sitting at the same abstraction level of CUDA C. You should be interested if you know (or want to learn) how to program a parallel accelerator like a GPU, while dealing with tricky performance characteristics and communication semantics.”
Native GPU programming with CUDAnative.jl
http://julialang.org/blog/2017/03/cudanative
“You can now write your CUDA kernels in Julia, albeit with some restrictions, making it possible to use Julia’s high-level language features to write high-performance GPU code.”
“The programming support we’re demonstrating here today consists of the low-level building blocks, sitting at the same abstraction level of CUDA C. You should be interested if you know (or want to learn) how to program a parallel accelerator like a GPU, while dealing with tricky performance characteristics and communication semantics.”