{"id":4769,"date":"2016-06-03T08:35:00","date_gmt":"2016-06-03T08:35:00","guid":{"rendered":"https:\/\/www.modernescpp.com\/index.php\/asynchronous-callable-wrappers\/"},"modified":"2023-06-26T12:55:48","modified_gmt":"2023-06-26T12:55:48","slug":"asynchronous-callable-wrappers","status":"publish","type":"post","link":"https:\/\/www.modernescpp.com\/index.php\/asynchronous-callable-wrappers\/","title":{"rendered":"Asynchronous Callable Wrappers"},"content":{"rendered":"

std::packaged_task<\/span> enables you to write a simple wrapper for a callable, which you can invoke later.<\/p>\n

<\/p>\n

std::packaged_task<\/h2>\n

To deal with std::packaged_task,<\/span> you have to perform four steps usually:<\/p>\n

    \n
  1. Wrap the task<\/li>\n
  2. Create the future<\/li>\n
  3. Perform the calculation<\/li>\n
  4. Pick up the result<\/li>\n<\/ol>\n

    These steps are easy to get with an example.<\/p>\n

    <\/p>\n

    \n\n\n\n
    \n
     1\r\n 2\r\n 3\r\n 4\r\n 5\r\n 6\r\n 7\r\n 8\r\n 9\r\n10\r\n11\r\n12\r\n13\r\n14\r\n15\r\n16\r\n17\r\n18\r\n19\r\n20\r\n21\r\n22\r\n23\r\n24\r\n25\r\n26\r\n27\r\n28\r\n29\r\n30\r\n31\r\n32\r\n33\r\n34\r\n35\r\n36\r\n37\r\n38\r\n39\r\n40\r\n41\r\n42\r\n43\r\n44\r\n45\r\n46\r\n47\r\n48\r\n49\r\n50\r\n51\r\n52\r\n53\r\n54\r\n55\r\n56\r\n57\r\n58\r\n59\r\n60\r\n61\r\n62\r\n63\r\n64\r\n65\r\n66\r\n67<\/pre>\n<\/td>\n
    		\n
    \/\/ packagedTask.cpp<\/span>\r\n\r\n#include <utility><\/span>\r\n#include <future><\/span>\r\n#include <iostream><\/span>\r\n#include <thread><\/span>\r\n#include <deque><\/span>\r\n\r\nclass<\/span> SumUp<\/span>{\r\n  public:\r\n    int<\/span> operator<\/span>()(int<\/span> beg, int<\/span> end){\r\n      long<\/span> long<\/span> int<\/span> sum{0};\r\n      for<\/span> (int<\/span> i= beg; i < end; ++i ) sum += i;\r\n      return<\/span> sum;\r\n    }\r\n};\r\n\r\nint<\/span> main(){\r\n\r\n  std::cout << std::endl;\r\n\r\n  SumUp sumUp1;\r\n  SumUp sumUp2;\r\n  SumUp sumUp3;\r\n  SumUp sumUp4;\r\n\r\n  \/\/ define the tasks<\/span>\r\n  std::packaged_task<int<\/span>(int<\/span>,int<\/span>)> sumTask1(sumUp1);\r\n  std::packaged_task<int<\/span>(int<\/span>,int<\/span>)> sumTask2(sumUp2);\r\n  std::packaged_task<int<\/span>(int<\/span>,int<\/span>)> sumTask3(sumUp3);\r\n  std::packaged_task<int<\/span>(int<\/span>,int<\/span>)> sumTask4(sumUp4);\r\n\r\n  \/\/ get the futures<\/span>\r\n  std::future<int<\/span>> sumResult1= sumTask1.get_future();\r\n  std::future<int<\/span>> sumResult2= sumTask2.get_future();\r\n  std::future<int<\/span>> sumResult3= sumTask3.get_future();\r\n  std::future<int<\/span>> sumResult4= sumTask4.get_future();\r\n\r\n  \/\/ push the tasks on the container<\/span>\r\n  std::deque< std::packaged_task<int<\/span>(int<\/span>,int<\/span>)> > allTasks;\r\n  allTasks.push_back(std::move(sumTask1));\r\n  allTasks.push_back(std::move(sumTask2));\r\n  allTasks.push_back(std::move(sumTask3));\r\n  allTasks.push_back(std::move(sumTask4));\r\n  \r\n  int<\/span> begin{1};\r\n  int<\/span> increment{2500};\r\n  int<\/span> end= begin + increment;\r\n\r\n  \/\/ execute each task in a separate thread<\/span>\r\n  while<\/span> ( not allTasks.empty() ){\r\n    std::packaged_task<int<\/span>(int<\/span>,int<\/span>)> myTask= std::move(allTasks.front());\r\n    allTasks.pop_front();\r\n    std::thread<\/span> sumThread(std::move(myTask),begin,end);\r\n    begin= end;\r\n    end += increment;\r\n    sumThread.detach();\r\n  }\r\n\r\n  \/\/ get the results<\/span>\r\n  auto<\/span> sum= sumResult1.get() + sumResult2.get() + sumResult3.get() + sumResult4.get();\r\n\r\n  std::cout << \"sum of 0 .. 10000 = \"<\/span> << sum << std::endl;\r\n\r\n  std::cout << std::endl;\r\n\r\n}\r\n<\/pre>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n
     <\/p>\n
    The job of the program is not so heavy. Calculate the sum from 0 to 10000 with the help of four threads and sum up the results with the associated futures. Of course, you can use the Gau\u00dfschen Summenformel<\/a>. (Strange, there is no English page describing this famous algorithm. But math is international.).<\/p>\n
    I pack the work packages in std::packaged_task (lines 28 – 31) objects in the first step. Work packages are instances of the class SumUp<\/span> (lines 9 – 16). The call operator does the current work (lines 11-15). The call operator sums up all numbers from beg<\/span> to end<\/span> and returns the sum<\/span>. std::packaged_task in lines 28 – 31 can deal with callables that need two ints and return an int.<\/p>\n
    Now, I have to create in the second step<\/strong> the future objects with the help of std::packaged_task<\/span> objects. Precisely that is done in lines 34 to 37. The packaged_task is the promise in the communication channel<\/a>. In these lines, I define the type of the future: std::future<int> sumResult1= sumTask1.get_future(<\/span>), but of course, I can do it also implicitly: auto sumResult1= sumTask1.get_future()<\/span>.<\/p>\n
    In the third step<\/strong>, the work takes place. The packaged_task is moved onto the std::deque<\/span> (lines 40 – 44). Each packaged_task (lines 51 – 58) is executed in the while loop. To do that, I move the head of the std::deque<\/span> in a std::packaged_task<\/span> (line 52), move the packaged_task in a new thread (line 54) and let it run in the background (line 56). std::packaged_task<\/span> object is not copyable. That’s why I used move semantic in lines 52 and 54. This restriction holds for all promises but also futures and threads. There is one exception to this rule: std:.shared_future.<\/p>\n
    In the fourth and last step<\/strong>, the four futures ask with the get call for the results and sum them up (line 61).<\/p>\n<\/p>\n
    Honestly, std::packaged_task<\/span> was not made for a simple use case like std::async<\/span>. But the result is simple.<\/p>\n
    \"packagedTask\"<\/p>\n
    Optimization potential<\/h3>\n
    C++11 has a function std::thread_hardware_concurrency<\/span>. It provides a hint about the number of cores on your system. If the C++ runtime has no glue, it conforms to the standard to return 0. So you should verify that value in your program (line 31). With the current GCC, Clang, or Microsoft compiler, I get the right answer 4. I use this number of cores In the variation of the program packagedTask.cpp<\/span>, to adjust the software to my hardware. So, my hardware is fully utilized.<\/p>\n
    <\/p>\n
    \n\n\n\n
    \n
     1\r\n 2\r\n 3\r\n 4\r\n 5\r\n 6\r\n 7\r\n 8\r\n 9\r\n10\r\n11\r\n12\r\n13\r\n14\r\n15\r\n16\r\n17\r\n18\r\n19\r\n20\r\n21\r\n22\r\n23\r\n24\r\n25\r\n26\r\n27\r\n28\r\n29\r\n30\r\n31\r\n32\r\n33\r\n34\r\n35\r\n36\r\n37\r\n38\r\n39\r\n40\r\n41\r\n42\r\n43\r\n44\r\n45\r\n46\r\n47\r\n48\r\n49\r\n50\r\n51\r\n52\r\n53\r\n54\r\n55\r\n56\r\n57\r\n58\r\n59\r\n60\r\n61\r\n62\r\n63\r\n64\r\n65\r\n66\r\n67\r\n68\r\n69\r\n70\r\n71\r\n72<\/pre>\n<\/td>\n
    		\n
    \/\/ packagedTaskHardwareConcurrency.cpp<\/span>\r\n\r\n#include <algorithm><\/span>\r\n#include <future><\/span>\r\n#include <iostream><\/span>\r\n#include <thread><\/span>\r\n#include <deque><\/span>\r\n#include <vector><\/span>\r\n\r\nclass<\/span> SumUp<\/span>{\r\npublic:\r\n  SumUp(int<\/span> b, int<\/span> e): beg(b),end(e){}\r\n  int<\/span> operator<\/span>()(){\r\n    long<\/span> long<\/span> int<\/span> sum{0};\r\n    for<\/span> (int<\/span> i= beg; i < end; ++i ) sum += i;\r\n    return<\/span> sum;\r\n  }\r\nprivate:\r\n    int<\/span> beg;\r\n    int<\/span> end;\r\n};\r\n\r\nstatic<\/span> const<\/span> unsigned<\/span> int<\/span> hwGuess= 4;\r\nstatic<\/span> const<\/span> unsigned<\/span> int<\/span> numbers= 10001;\r\n\r\nint<\/span> main(){\r\n\r\n  std::cout << std::endl;\r\n\r\n  unsigned<\/span> int<\/span> hw= std::thread<\/span>::hardware_concurrency();\r\n  unsigned<\/span> int<\/span> hwConcurr= (hw != 0)? hw : hwGuess;\r\n\r\n  \/\/ define the functors<\/span>\r\n  std::vector<SumUp> sumUp;\r\n  for<\/span> ( unsigned<\/span> int<\/span> i= 0; i < hwConcurr; ++i){\r\n    int<\/span> begin= (i*numbers)\/hwConcurr;\r\n    int<\/span> end= (i+1)*numbers\/hwConcurr;\r\n    sumUp.push_back(SumUp(begin ,end));\r\n  }\r\n\r\n  \/\/ define the tasks<\/span>\r\n  std::deque<std::packaged_task<int<\/span>()>> sumTask;\r\n  for<\/span> ( unsigned<\/span> int<\/span> i= 0; i < hwConcurr; ++i){\r\n    std::packaged_task<int<\/span>()> SumTask(sumUp[i]);\r\n    sumTask.push_back(std::move(SumTask));\r\n  }\r\n\r\n  \/\/ get the futures<\/span>\r\n  std::vector< std::future<int<\/span>>> sumResult;\r\n  for<\/span> ( unsigned<\/span> int<\/span> i= 0; i < hwConcurr; ++i){\r\n    sumResult.push_back(sumTask[i].get_future());\r\n  }\r\n\r\n  \/\/ execute each task in a separate thread<\/span>\r\n  while<\/span> ( not sumTask.empty() ){\r\n    std::packaged_task<int<\/span>()> myTask= std::move(sumTask.front());\r\n    sumTask.pop_front();\r\n    std::thread<\/span> sumThread(std::move(myTask));\r\n    sumThread.detach();\r\n  }\r\n\r\n  \/\/ get the results<\/span>\r\n  int<\/span> sum= 0;\r\n  for<\/span> ( unsigned<\/span> int<\/span> i= 0; i < hwConcurr; ++i){\r\n    sum += sumResult[i].get();\r\n  }\r\n\r\n  std::cout << \"sum of 0 .. 100000 = \"<\/span> << sum << std::endl;\r\n\r\n  std::cout << std::endl;\r\n\r\n}\r\n<\/pre>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n
     <\/p>\n
    What’s next?<\/h2>\n
    The next post <\/a>will be about futures and promises. (Proofreader Alexey Elymanov<\/strong>)<\/p>\n
     <\/p>\n
     <\/p>\n
     <\/p>\n
     <\/p>\n
     <\/p>\n
     <\/p>\n
     <\/p>\n","protected":false},"excerpt":{"rendered":"
    std::packaged_task enables you to write a simple wrapper for a callable, which you can invoke later.<\/p>\n","protected":false},"author":21,"featured_media":4768,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[366],"tags":[446],"class_list":["post-4769","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-multithreading","tag-tasks"],"_links":{"self":[{"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/posts\/4769","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/users\/21"}],"replies":[{"embeddable":true,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/comments?post=4769"}],"version-history":[{"count":1,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/posts\/4769\/revisions"}],"predecessor-version":[{"id":6976,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/posts\/4769\/revisions\/6976"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/media\/4768"}],"wp:attachment":[{"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/media?parent=4769"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/categories?post=4769"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/tags?post=4769"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}