Dynamic distributed execution for data parallel tasks in Julia

I was assigned a homework.
Task : Dynamic distributed execution for data parallel tasks in Julia.

This project proposal is to implement a native Julia framework for distributed execution for general purpose data parallelism, using dynamic, runtime-generated general task graphs which are flexible enough to describe multiple classes of parallel algorithms. You will be expected to weave together native Julia parallelism constructs such as the ClusterManager for massively parallel execution, and automate the handling of data dependencies using native Julia RemoteRefs as remote data futures and handles. You will also be encouraged to experiment with novel scheduling algorithms.

This project is from http://www_old.julialang.org/soc/projects/hpc.html
But I have no idea.
Welcome everyone’s advice.

Implemented already:

although there’s plenty of room for improvements, possibly around adding more kinds of scheduling algorithms.

And if I may ask, who is assigning you Julia GSoC ideas as homework??? :stuck_out_tongue:

1 Like