03-10-2020 11:05 AM
@mcduff wrote:
Besides complex numbers I thought you liked to do things in parallel.
Well, addition is such a simple operation and all values depend on other values. Even if you would parallelize (or optimized for SSE, etc.) , the partial results will still need to be added later in one way or another, just increasing the workload.
Most code discussed here touches every element exactly once and there is one ADD per element, so I doubt much improvements are possible (O(N)). My code is extremely lightweight (buffers or not), so most likely all other other computations involving these large data structures are orders of magnitude more impactful on overall performance.
This is probably not the place to dig deep, because it does not really matter. ;). Create a subVI, disable debugging, and inline it. Voila! 😄