[Python-Dev] Performance compares

[Tim Peters]

[MAL]
> Since it is possible that these figures result from my specific
> machine setup, I'd like to know what other people see on their
> machines.

Is this the same machine where you were able to get 15% difference a few
years ago by adding or removing an unreachable printf in ceval.c (or was that
Vladimir)?  If so, I bet it's degenerated to random 50% difference since then
<wink>.

My Win98SE box is *astonishingly* useless for timings.  Without fail, the
first time I run pystone after a reboot yields a result a solid 50% higher
than the second or subsequent times I run it (yes, it's major-league *slower*
the second time).  This is true across dozens of trials over several months,
and across all versions of Python.

And simple little loops routinely vary in reported runtime by a factor of 3.
I may have to dig my old Win95 box out of the packing crate <0.6 wink>.

None of that changes, of course, that the numbers you got are scary.