Why doesn't JUMP_IF_FALSE do POP_TOP ?

Jeff Epler jepler at unpythonic.net
Mon Jan 13 18:30:01 CET 2003


> On Mon, 13 Jan 2003 10:48:30 -0500, "Terry Reedy" <tjreedy at udel.edu> wrote:

> >When I asked about this in August one answer was that this should not
> >be changed until the 'new compiler' is finished (it still is not)
> >because adding an opcode would create foward/backward incompatibilites
> >that are not justified by such a minor optimization.
> >
On Mon, Jan 13, 2003 at 05:04:30PM +0000, Bengt Richter wrote:
> Has anyone taken a measure of exactly how "minor" it really would be?

I know this has been written long ago.  A person did a dynamic analysis
of instruction pair frequency, added them as separate VM instructions,
and benchmarked the result for performance improvement.

I can't find the paper now.  Maybe the paper I described hasn't been
written, and this is the one I was thinking of:
http://www.foretec.com/python/workshops/1998-11/proceedings/papers/aycock-211/aycock211.html

Jeff





More information about the Python-list mailing list