multithreading the algorithm

[am2222]

Hi,
can we increase this algorithm's speed by multithreading?is it possible even?

[jonnyhuck]

I would have thought so, you could split the circle into two halves, or quarters etc and run each one in a different thread, though I do wonnder if there would be much to gain given the speed that it already runs at - it might possibly be useful where there was either a very large viewshed or very high resolution data, for example.

The only thing to be careful of would be to make sure that you don't get any disagreements along the joins between the sections, and that these are resolved (e.g. if a boundary cell was classified as visible in one section and invisible in another).

It's definitely worth having a go at if you fancy it though, it would be interesting to see what difference it makes!

[am2222]

@jonnyhuck thanks for your reply, what about calculating each beam in on thread? As we loop 360 degree in each beam we define a separate thread? I think this will not face any borders problem.

[jonnyhuck]

You could certainly give it a go! I suspect that there might come a point where the 'law of diminishing returns' kicks in, but I would love to hear how it goes!

[am2222]

@jonnyhuck hi,
I tried to use qtconcurrent to make it parallel. but faced memory lakes. can you please help me about it? I could not find its source
here is my question
https://stackoverflow.com/questions/48130979/invalid-read-of-size-4-in-valgrind-test
thanks