In this article we will discuss how to execute tasks asynchronously with std::async in C++11.
std::async is introduced in c++11.
what is std::async()
std::async() is a function template that accepts a callback(i.e. function or function object) as an argument and potentially executes them asynchronously.
template <class Fn, class... Args> future<typename result_of<Fn(Args...)>::type> async (launch policy, Fn&& fn, Args&&... args);
std::async returns a std::future<T>, that stores the value returned by function object executed by std::async(). Arguments expected by function can be passed to std::async() as arguments after the function pointer argument.
First argument in std::async is launch policy, it control the asynchronous behaviour of std::async. We can create std::async with 3 different launch policies i.e.
Frequently Asked:
- std::launch::async
- It guarantees the asynchronous behaviour i.e. passed function will be executed in seperate thread.
- std::launch::deferred
- Non asynchronous behaviour i.e. Function will be called when other thread will call get() on future to access the shared state.
- std::launch::async | std::launch::deferred
- Its the default behaviour. With this launch policy it can run asynchronously or not depending on the load on system. But we have no control over it.
If we do not specify an launch policy. Its behaviour will be similar to std::launch::async | std::launch::deferred.
We are going to use std::launch::async launch policy in this article.
We can pass any callback in std::async i.e.
- Function Pointer
- Function Object
- Lambda Function
Let’s understand the need of std::async by an example,
Best Resources to Learn C++:
Need of std::async()
Suppose we have to fetch some data (string) from DB and some from files in file-system. Then I need to merge both the strings and print.
In a single thread we will do like this,
#include <iostream> #include <string> #include <chrono> #include <thread> using namespace std::chrono; std::string fetchDataFromDB(std::string recvdData) { // Make sure that function takes 5 seconds to complete std::this_thread::sleep_for(seconds(5)); //Do stuff like creating DB Connection and fetching Data return "DB_" + recvdData; } std::string fetchDataFromFile(std::string recvdData) { // Make sure that function takes 5 seconds to complete std::this_thread::sleep_for(seconds(5)); //Do stuff like fetching Data File return "File_" + recvdData; } int main() { // Get Start Time system_clock::time_point start = system_clock::now(); //Fetch Data from DB std::string dbData = fetchDataFromDB("Data"); //Fetch Data from File std::string fileData = fetchDataFromFile("Data"); // Get End Time auto end = system_clock::now(); auto diff = duration_cast < std::chrono::seconds > (end - start).count(); std::cout << "Total Time Taken = " << diff << " Seconds" << std::endl; //Combine The Data std::string data = dbData + " :: " + fileData; //Printing the combined Data std::cout << "Data = " << data << std::endl; return 0; }
Output:
Total Time Taken = 10 Seconds Data = DB_Data :: File_Data
As both the functions fetchDataFromDB() & fetchDataFromFile() takes 5 seconds each and are running in a single thread so, total time consumed will be 10 seconds.
Now as fetching data from DB and file are independent of each other and also time consuming. So, we can run them in parallel.
One way to do is create a new thread pass a promise as an argument to thread function and fetch data from associated std::future object in calling thread.
The other easy way is using std::async.
Calling std::async with function pointer as callback
Now let’s modify the above code and call fetchDataFromDB() asynchronously using std::async() i.e.
std::future<std::string> resultFromDB = std::async(std::launch::async, fetchDataFromDB, "Data"); // Do Some Stuff //Fetch Data from DB // Will block till data is available in future<std::string> object. std::string dbData = resultFromDB.get();
std::async() does following things,
- It automatically creates a thread (Or picks from internal thread pool) and a promise object for us.
- Then passes the std::promise object to thread function and returns the associated std::future object.
- When our passed argument function exits then its value will be set in this promise object, so eventually return value will be available in std::future object.
Now change the above example and use std::async to read data from DB asyncronously i.e.
Checkout the compete example as follows,
#include <iostream> #include <string> #include <chrono> #include <thread> #include <future> using namespace std::chrono; std::string fetchDataFromDB(std::string recvdData) { // Make sure that function takes 5 seconds to complete std::this_thread::sleep_for(seconds(5)); //Do stuff like creating DB Connection and fetching Data return "DB_" + recvdData; } std::string fetchDataFromFile(std::string recvdData) { // Make sure that function takes 5 seconds to complete std::this_thread::sleep_for(seconds(5)); //Do stuff like fetching Data File return "File_" + recvdData; } int main() { // Get Start Time system_clock::time_point start = system_clock::now(); std::future<std::string> resultFromDB = std::async(std::launch::async, fetchDataFromDB, "Data"); //Fetch Data from File std::string fileData = fetchDataFromFile("Data"); //Fetch Data from DB // Will block till data is available in future<std::string> object. std::string dbData = resultFromDB.get(); // Get End Time auto end = system_clock::now(); auto diff = duration_cast < std::chrono::seconds > (end - start).count(); std::cout << "Total Time Taken = " << diff << " Seconds" << std::endl; //Combine The Data std::string data = dbData + " :: " + fileData; //Printing the combined Data std::cout << "Data = " << data << std::endl; return 0; }
Now it will take 5 seconds only.
Output:
Total Time Taken = 5 Seconds Data = DB_Data :: File_Data
Calling std::async with Function Object as callback
/* * Function Object */ struct DataFetcher { std::string operator()(std::string recvdData) { // Make sure that function takes 5 seconds to complete std::this_thread::sleep_for (seconds(5)); //Do stuff like fetching Data File return "File_" + recvdData; } }; //Calling std::async with function object std::future<std::string> fileResult = std::async(DataFetcher(), "Data");
Calling std::async with Lambda function as callback
//Calling std::async with lambda function std::future<std::string> resultFromDB = std::async([](std::string recvdData){ std::this_thread::sleep_for (seconds(5)); //Do stuff like creating DB Connection and fetching Data return "DB_" + recvdData; }, "Data");
The tutorial is cool like an Ice cream…very easy to understand and enjoy the new features of threading without any confusion and waste of time…Thank You, Varun
Your tutorials are amazing!, all this articles are my main reference for C++ threads. Thank you!
Nice tutorials. Thanks for sharing..
Could you please clang-format C++ code in thread related tutorials? Code would be much easier to read. If you have Github repo, I can make pull request, if you want.
Great work. thanks for nice explanation.
Amazing articles. Thanks so much