Note that this package was deprecated in favor of python chunkflow.

the computation was defined as a node. The kinds of node was defined in a task file

julia main.jl --help
julia main.jl -h

julia main.jl --task=task.json --deviceid=0
julia main.jl -t task.json -d 0

• define task.json file, the file name can be only a prefix. Task producer will match all the files and produce a bunch of tasks
• produce tasks: julia taskproducer.jl --task=task.json --awssqs=chunkflow-tasks

find help:

julia taskproducer.jl -h

launch a number of instances to deal with the tasks. run the same commands in each instance

julia -j number_of_processes main.jl -n number_of_processes -w wait_seconds_of_each_launch -q queue-name