>Just a general comment, without looking at the specifics of your tests whatsoever: sometimes when speed testing I've seen large drops in performance once a test overwhelms a CPU's L1 cache. If one of your tests is overwhelming the cache, and the other isn't, that could explain otherwise anomalous results.

Good point, Al, and I do not think that is the case here

But just in case, I'll redo the test on my normal computer once it's repaired - I'm on an older notebook here for the moment