I was poking around the
VCL system.math
unit yesterday when I came across a curious function implementation. Not curious because of complexity, but rather baffling as to why they would implement it like this. While the performance gain in question is insignificant in the great scope of things, the backstory here is that the
VCL is full of similarly written code. Code that, when you sum up the penalty on application level, becomes quite considerable.
Weiterlesen...