Does anyone know how I can assign to 2 processors on a dual core computer for the combination of Simpson's rule (y[n+2]-y[n]=(h/3)*(f[n+2]+4f[n+1]+f[n])) and two-step Adams Moulton method (y[n+2]-y[n+1]=(h/12)*(5f[n+2]+8f[n+1]-f[n])) to solve ordinary differential equations to produce results for y[n+1] and y[n+2] simultaneously from each processor using Maple, Matlab or Mathematica?
I guess you have an self-pedagogical ambition with that. If you are simply interested in the result, you will be much faster by simply programming in C or C++ and let your system decide how to make use of the two cores.
There is a very easy way to use multicore plataforms in the newer matlab releases.
If you browse the paralel menu in the main window, you can find several ways to use the application. To use in a full way the cores you do add a few lines in the code and that is all.
Dear Ehigie,
I agree with Ulrich that if you are interested in the result it is much simpler to implement the algorithm in, e.g., C with separate threads. Then, these threads can be bound to specific cores by using cpu affinity. For instance, Linux OS allows you to directly set the affinity by invoking 'sched_setaffinity' while in windows you can do a similar operation by invoking 'SetThreadAffinityMask'.
You can use matlab parallel for loop (parfor) for that. Also you can consider mixing matlab with c (C_MEX) and/or CUDA.
- Badges
- Science topic