Description
Scientific code that must run on NVIDIA, AMD, Intel and Apple GPUs usually needs a version per vendor. JACC.jl lets Julia users write one high-level code that runs in parallel on any CPU or GPU, like Kokkos in C++.
It offers unified array allocation plus parallel_for and parallel_reduce, funded by the US DOE.
One code:Any hardware.
Unified arrays:Backend chosen for you.
Primitives:for and reduce.
It offers unified array allocation plus parallel_for and parallel_reduce, funded by the US DOE.
Features
One code:Any hardware.
Unified arrays:Backend chosen for you.
Primitives:for and reduce.
Screenshots
Tags:gpu