Interesting though this is, it does bring up something I recall from (more than) a few years back. I cut my coding teeth on a Commodore, and remember an experiment showing the difference between coding and compiling with an ultimately simple program that changed the colour of the screen through the basic range of 16 colours. Written in basic and then compiled into machine code, it took about 10 seconds to go through the range. The same exercise done directly in source code, altered the screen faster than the eye could see.
Is there a similar slowdown in compiling source code for Linux? It is my understanding that this is one of the reasons why Windows software needs such gigantic computing power for relatively simple tasks; too much redundant code from the compiler to make it all work.

