Implicitly-parallelizing programming - Flow Manifesto(flowlang.net)
flowlang.net
Implicitly-parallelizing programming - Flow Manifesto
http://www.flowlang.net/p/flow-manifesto.html
3 comments
>"Write once, parallelize anywhere" -- With very little modification, the same Flow code should be able to target hadoop / Google's MapReduce, the JVM using Java Threads, C using pthreads, MPI or a GPU using CUDA. By understanding the big-Oh complexity of computations in Flow programs, The compiler will produce code that optimally partitions the workload in systems with nonuniform computing speeds and/or communication costs, e.g. CPU/GPU combinations.
That's a big promise.
That's a big promise.
When I read through the text, it didn't once think "this can't work". It seems plausible and if implemented, this could indeed be a big step forward.
I hope this, or a similar effort, will be actually be realized.
I hope this, or a similar effort, will be actually be realized.
Call me old if you like, but I've seen countless of similar "manifestos" covering everything from flight simulators to, yes, programming languages. Only on very rare occasions do they result in something concrete.