Yeah, I didn't study the code very closely. I was thinking the test got 'bigger' when it really is just repeated more times.
That slowdown of almost exactly 2X sure is suspicious. It's what made me think the explanation must be a matter of primary cache versus secondary ( half speed ) cache.
That slowdown of almost exactly 2X sure is suspicious. It's what made me think the explanation must be a matter of primary cache versus secondary ( half speed ) cache.