04-12-2011 02:11 PM
Hi-
I am trying to develop LV code (Windows OS) to perform pipelined FFT on a 2D input array.
The code has to programmatically scale with the number of processor cores available.
Specifically, If N cores are present the code will:
1) Split the array into N equal parts
2) Run N FFTs simultaneously
3) Merge N processed arrays back into 1 array
And N can change depending on the specific processor used.
Any suggestions?
Best,
Dar Bahatt
04-12-2011 02:22 PM
Have you checked out the High Performance Analysis Libraries?
04-13-2011 03:03 PM
You may look at the For Loop Iteration Parallelism, which allows you to specify certain tasks among multipl processors, and run them in parallel.