3 ms·
There is also the embedded functional language futhark [0], which compiles numerical kernels to optimized c, ispc, opencl and cuda code. Parallelism is explicit
by beyondCritics 4y ago
There is also the embedded functional language futhark [0], which compiles numerical kernels to optimized c, ispc, opencl and cuda code. Parallelism is explicit using known constructs (map, reduce,...).
Documentation is excellent and there is also a SML style type system, hence you can rely on strong abstractions.
[0] https://futhark-lang.org/ https://futhark-lang.org/