How to accelerate matrix operations(multiplication, add, inverse) in a for loop?

I think you are basically right in your remarks, but we should avoid creating a misunderstanding here. Julia can do slices and vectorization etc. etc. just fine. But in Matlab you do it because you have to. In Julia you don’t, and you can do much better.

2 posts were split to a new topic: Julia vs Matlab side thread

Thank you so much!

Hi all, I cleaned up the code a little bit, it turns out the bottleneck is at this line:

corr = (noisyPatch[px,py] ⋅ b) / noisyPatchNorm[px,py]

any idea on how to optimize this?

1 Like