About the parallel solving by multiple Thread？

[jh chen]

Hi, ，guys！ Current I try to solve problems like the following：

1. get a dateset，for example, 4000 individuals

2. split the dataset into 4 pieces, d1,d2 ,d3,d4

3.initalize solving contexts: c1,c2,c3,c4

4. initialize a thread pool "mypool" with 2 core threads, new a semaphore "semaphore"with 2 permits.

5. a for loop to start each thread, 

for i in contexts.sized():

   mythread->run t{

semaphore.acquire();

new solving(ci,seamphore);//in here, i try use the same solver, and different solver.

                                                //new a completely different sover for each thread is a favorable way

}

solving(ci,semaphore){

.........

this.solvemanger(ci)

semaphore.release()

}

----------------------------------------------------------------------------------------------------------------------------------------------

From my description, you can know that I want to solve the problem by multi-thread, but something wrong happened.

when the semaphore=1. core thread=1: c1 finished, c2 initialized, but didn't continue. i.e., c2can get an initial score, and then nothing happens, everything stops. it seems like all done.

when the semaphore=2. core thread=2: c1,c2 finished, c3,c4 initialized, but didn't continue. i.e., c3, c4 can get an initial score, and then nothing happens, everything stops. it seems like all done.

after the above situation appears, if you send a request, it will run into a new same procedure like the above. is anything wrong about my idea？

-------------------------------------------------------------------------------------------------------------------------------------------

I found in the manual of optaplanner that “Multitenancy: solve different datasets in parallel： The SolverManager will make it even easier to set this up, in a future version.”

I would like to ask you sirs, on this version, if I want to solve different data sets in parallel, how to set it up.

Or, could such an idea be implemented in the current version?