parallel computing-- sharing viable between threads

Hi, recently, I am working on a project that needs parallel computing. I already know things like parfor. However, the difficulty in this project is that, it requires sharing a global variable between threads. For example, if there are two threads both computing "a=a+1;"in a for loop, than the parfor results should be "lab1 : 1,2,3,4,5,6...; lab2 : 1,2,3,4,5,6...", what I expect is probably like this:"lab1: 1,3,6,7,9... lab2: 2,4,5,8,10...." which two threads are both changing the same variable (It doesn't care the sequence of which thread does the add, as long as all threads are adding the number together) Can anybody help me with that?

Answer 1

[Walter Roberson]

It is not possible to have a global shared between workers.

You can use spmd to labSend and labReceive

As of the newly released r2017a there is the ability to send between workers. I would need to check the details.

You might be able to benefit from the sharedmatrix file exchange contribution

[yu liu]

Hi, Roberson. Thanks for reply, I'll check that

[Walter Roberson]

The new facilities in R2017a are send(), poll(), and afterEach(). However, these facilities are to transfer data back and forth between the client and the workers, and are not for transferring data between workers: to transfer between workers, spmd with labSend() and labReceive() is still the mechanism.