github-starred/ultimatepp
2
0
Fork 0
mirror of https://github.com/ultimatepp/ultimatepp.git synced 2026-05-15 14:16:07 -06:00
Code Issues 33 Projects Packages Wiki Activity Actions

.tutorial

Browse source 
git-svn-id: svn://ultimatepp.org/upp/trunk@13585 f0d560ea-af0d-0410-9eb7-867de7ffcac7
This commit is contained in:
cxl 2019-09-05 12:53:05 +00:00
parent b6419836e7
commit 57201c576a
1 changed files with 13 additions and 13 deletions
Download patch file Download diff file Expand all files Collapse all files

26
tutorial/CoreTutorial/CoLoop.cpp
View file

@ -2,30 +2,30 @@
void CoLoopTutorial()
{
/// .CoWork loop method
/// .CoDo
/// An alternative to `CoPartition` is 'loop' method of `CoWork` (`Loop` method with synonym
/// `operator*`). In this pattern, the job is simply started in all threads and the code is
/// responsible for scheduling the work. `Loop` / `operator*` waits for all started threads
/// to finish. For scheduling, it is possible to use internal `CoWork` index counter,
/// which is set to 0 at the start of `Loop` / `operator*`. This way, the overhead associated
/// with creating lambdas and scheduling them is kept to the minimum.
/// An alternative to `CoPartition` is `CoDo`. In this pattern, the job is simply started
/// in all threads and the code is responsible for scheduling the work. `CoDo` waits for
/// all started threads to finish. Scheduling is the responsibility of client code, but can
/// be easily managed using the std::atomic counter. This way, the overhead associated with
/// creating lambdas and scheduling them is kept to the minimum (basically the cost of
/// atomic increment).
Vector<String> data;
for(int i = 0; i < 100; i++)
data.Add(AsString(1.0 / i));
double sum = 0;
CoWork co;
co * [&] {
int i;
std::atomic<int> ii = 0;
CoDo([&] {
double m = 0;
while((i = co.Next()) < data.GetCount())
for(int i = ii++; i < data.GetCount(); i = ii++)
m += atof(data[i]);
CoWork::FinLock();
sum += m;
};
});
DUMP(sum);