Discrepancy in measuring performance?

[William E. Skeith III]

Sorry for generating so much traffic in one day : (  I have (sort of)
figured out what the issue is.  It turns out I am affected by a bug in
the linux kernel (or perhaps just my distribution):

https://bugs.archlinux.org/task/28859

I ran powertop, and noticed *tons* of interrupts from acpi.  After
booting with acpi disabled, I got exactly the results I expected from my
test program.  Sorry for wasting everyone's time.

-WES

p.s. I'm now extra impressed by the speed program for finding a way to
make itself immune to this.