One block of memory with two pointers for two different processes in unmanaged situation?

[MyCatAlex]

I have a problem of transferring some integer values that are generated at a fast clip in MFCaptureD3D. The latter is a GitHub project that captures pixel values of standard WebCam. It is a C++ application. A few variants have been considered, including named pipes and named shared memory. They still might be used but analyzing the code I noticed that there is considerable overhead in each method and that I am afraid will make them impractically slow and chocked with the data. So I want to define one block of memory with the stipulation that I could also define two pointers attached to it. I think I can use one of them to write the data and the other one to read. My question is are there any pitfalls in the idea? If there are I will abandon it and try to do named pipes.

int* ptr1[76800];

So, how do I define the second pointer if it is at all possible? process A, writer ==> [memory] ==> process B, reader.

Thanks, -MyCatAlex

Continue reading...