Thank you so much for the vector information! I've found it very difficult to find good resources on vectors in a high-performance environment, so I really appreciate your input. In my experience, Boost tends to slow things down enough that I'm okay with programming the deletion myself. It'll be good practice, anyway.
I really like the idea of passing a data chunk into a point constructor. That sounds very clean, especially since the file header also gives me the size-on-disk of each point format, I already know programmatically how big of a chunk to read for each point.
Since my data file is stored in a binary format (not ASCII text), would it be faster to pull data one point (30 bytes, for example) at a time, or one component at a time (2 bytes)? I imagine it would be the former because of the reduced number of calls to the disk, especially with a typical HDD. In that case, would it make sense to read in a QByteArray using QIODevice::read(qint64) and pass that to the point constructor?
I desperately need to take a good course on code optimization so I don't have to rely on you all. Thanks again! This should be my last question.
Bookmarks