splitting for loop into separate threads to increase CPU usage

[dudewhatsup]

Hello,

in particular my task is to accelerate saving an AVI file with OpenCV. Inside the for loop each pixel of the current frame is altered and then put into a new image (via cvSet2D), then saved via cvWriteFrame to the AVI. I thought that if I let one thread process every 2nd pixel and let the other thread do the same - being 1 pixel ahead - that this could possibly increase saving speed. So I would have two threads, i.e. two for-loops with +=2 increments.

The point is, so far CPU usage was around 50% (dual core) during the saving process and I tried programming two threads (with _beginthread) with IplImages inside the void functions but somehow it won't work as 1. I'm a newbie in regard to threads and 2. I don't know if it's even feasible.

Any suggestions how to accomplish that task or would it be too complicated for a newbie. Perhaps there's 'just a few lines solution' ?

Best
dude

[htirwin]

Can you post the loop? Maybe you can use #pragma omp parallel for ?

[dudewhatsup]

I've tried OpenMP already and I got a message that told me that some test for the loop was not successful, i.e. the for loop didn't meet the requirements for that command.