[Python-Dev] Re: Python 2.1 slower than 2.0

Jeremy Hylton jeremy@alum.mit.edu
Mon, 29 Jan 2001 11:48:11 -0500 (EST)

>>>>> "MAL" == M -A Lemburg <mal@lemburg.com> writes:

  MAL> Yes... after a check of the Makefile I found that I had
  MAL> compiled Python 2.0 with -O3 and 2.1a1 with -O2 -- perhaps this
  MAL> makes a difference w/r to inlining of code. I'll recompile and
  MAL> rerun the benchmark.
When I was working in the CALL_FUNCTION revision, I compared 2.0 final
with my development working using -O3.  At that time, I saw no
significant performance difference between the two.  And I did notice
a difference between -O2 and -O3.

The strange thing is that I notice a difference between -O2 and -O3
with 2.1a1, but in the opposite direction.  On pystone, python -O2
runs consistently faster than -O3; the difference is .05 sec on my