Taskflow Algorithms » Module Algorithm

Contents

• Include the Header
• What is a Module Task
• Create a Module Task over a Custom Graph

Taskflow provides template methods that let users create reusable building blocks called modules. Users can connect modules together to build more complex parallel algorithms.

Include the Header

You need to include the header file, taskflow/algorithm/module.hpp, for creating a module task over a schedulable graph target.

#include <taskflow/algorithm/module.hpp>

What is a Module Task

Similar to Composable Tasking, but in a more generic setting, the template function, tf::make_module_task, allows you to create a task over a taskflow graph that can be executed by an executor. This provides a flexible way to encapsulate and reuse complex task logic within your Taskflow applications. The example below demonstrates how to create and launch multiple taskflows in parallel using asynchronous tasking:

tf::Executor executor;

tf::Taskflow A;
tf::Taskflow B;
tf::Taskflow C;
tf::Taskflow D;

A.emplace([](){ printf("Taskflow A\n"); }); 
B.emplace([](){ printf("Taskflow B\n"); }); 
C.emplace([](){ printf("Taskflow C\n"); }); 
D.emplace([](){ printf("Taskflow D\n"); }); 

// launch the four taskflows using asynchronous tasking
executor.async(tf::make_module_task(A));
executor.async(tf::make_module_task(B));
executor.async(tf::make_module_task(C));
executor.async(tf::make_module_task(D));
executor.wait_for_all();  

Since the four taskflows are launched asynchronously without any dependencies between them, we can observe any order of the output message:

# one possible output
Taskflow B
Taskflow C
Taskflow A
Taskflow D

# another possible output
Taskflow D
Taskflow A
Taskflow B
Taskflow C

If you need to enforce dependencies among these four taskflows, you can use dependent-async tasks. The example below launches the four taskflows one by one in sequential:

tf::Executor executor;

tf::Taskflow A;
tf::Taskflow B;
tf::Taskflow C;
tf::Taskflow D;

A.emplace([](){ printf("Taskflow A\n"); }); 
B.emplace([](){ printf("Taskflow B\n"); }); 
C.emplace([](){ printf("Taskflow C\n"); }); 
D.emplace([](){ printf("Taskflow D\n"); }); 

auto TA = executor.silent_dependent_async(tf::make_module_task(A));
auto TB = executor.silent_dependent_async(tf::make_module_task(B), TA);
auto TC = executor.silent_dependent_async(tf::make_module_task(C), TB);
auto [TD, FD] = executor.dependent_async(tf::make_module_task(D), TC);
FD.get();

# dependent-async tasks enforce a sequential execution of the four taskflows
Taskflow A
Taskflow B
Taskflow C
Taskflow D

The module task maker, tf::make_module_task, functions basically similar to tf::Taskflow::composed_of but provides a more generic interface that can be used beyond Taskflow. Specifically, the following two approaches achieve the same functionality.

// approach 1: composition using composed_of
tf::Task m1 = taskflow1.composed_of(taskflow2);

// approach 2: composition using make_module_task
tf::Task m1 = taskflow1.emplace(tf::make_module_task(taskflow2));

Attention

Similar to tf::Taskflow::composed_of, tf::make_module_task does not assume ownership of the provided taskflow but a soft reference. You are responsible for ensuring that the encapsulated taskflow remains valid throughout its execution.

Create a Module Task over a Custom Graph

In addition to encapsulate taskflow graphs, you can create a module task to schedule a custom graph target. A schedulable target (of type T) must define the method T::graph() that returns a reference to the tf::Graph object managed by T. The following example defines a custom graph that can be scheduled through making module tasks:

struct CustomGraph {
  tf::Graph graph;
  CustomGraph() {
    // use flow builder to inherit all task creation methods in tf::Taskflow
    tf::FlowBuilder builder(graph);
    tf::Task task = builder.emplace([](){
      std::cout << "a task\n";  // static task
    });
  }
  // returns a reference to the graph for taskflow composition
  Graph& graph() { return graph; }
};

CustomGraph target;
executor.async(tf::make_module_task(target));

Attention

Users are responsible for ensuring the given custom graph remains valid throughout its execution. The executor does not assume ownership of the custom graph.