[issue7593] Computed-goto patch for RE engine

[A.M. Kuchling]

Changes by A.M. Kuchling :


--
resolution:  -> wont fix
status: open -> closed

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue7593] Computed-goto patch for RE engine

[Ezio Melotti]

Changes by Ezio Melotti :


--
versions: +Python 3.4 -Python 3.2

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue7593] Computed-goto patch for RE engine

[Chris Leary]

Changes by Chris Leary :


--
nosy: +cdleary

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue7593] Computed-goto patch for RE engine

[Antoine Pitrou]

Antoine Pitrou  added the comment:

You should disassemble the output (or produce assembler from gcc) and check 
that the various indirect jumps at the end of each case block don't get merged 
into a single shared indirect jump.

Or perhaps it's simply that regular expression matching isn't really sensitive 
to bytecode dispatch overhead.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue7593] Computed-goto patch for RE engine

[A.M. Kuchling]

A.M. Kuchling  added the comment:

Actually, I really want someone to verify that measurement.  As a control, I 
tried running the call_method benchmark (after a few more xrange fixes).  The 
Python 3.x trunk version with my patch is measured as 1.0227x slower, even 
though the patch only touches the re module and call_method doesn't use the 
module at all.  I recompiled both binaries; both builds are using the same 
compiler arguments; both have the same version from trunk.  I'm mystified about 
why the patched version is slower.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue7593] Computed-goto patch for RE engine

[A.M. Kuchling]

A.M. Kuchling  added the comment:

I finally got around to benchmarking this change, and unfortunately the results 
are not good.

I used the regex tests in the Unladen Swallow test suite, regex_effbot and 
regex_v8.  The tests are written for Python 2.x, but the fixes for 3.x are 
straightforward (use print() in one function; replace xrange with range in the 
bm_regex_effbot.py and bm_regex_v8.py files; remove a few uses of u''; I'll 
provide a patch later.)

Hardware: MacBook, Intel Core 2 Duo, 1.83GHz, 2MB L2 cache, 667 MHz bus.

Tests invoked with ./perf.py -b regex_effbot -r -v ../py3k/python.exe 
../threaded-3000/python.exe

regex_effbot is 1.1002 times slower with the computed-goto patch, and 
regex_v8 is 1.0081x slower.  perf.py indicates that both results are 
significant.

I'd like to see a few people replicate these results -- maybe the effect is 
very platform dependent or I ran the tests incorrectly -- but on current 
evidence, this patch is not worth pursuing.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue7593] Computed-goto patch for RE engine

[Georg Brandl]

Georg Brandl  added the comment:

_sre is listed in Modules/Setup, so it will be a built-in module by default.

--
nosy: +georg.brandl

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue7593] Computed-goto patch for RE engine

[Antoine Pitrou]

Antoine Pitrou  added the comment:

On the principle nothing looks wrong. There are some tabs-vs-spaces
issues in _sre.c. Can some out-of-bounds crash be triggered if an opcode
is greater than 33?

It needs some benchmarks to know whether it's efficient. Also, I think
it would be nice to include regeneration of the opcode table in the
build process. Since it's very light-weight it can be done
unconditionally (I suppose it will be done from setup.py, right?).

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue7593] Computed-goto patch for RE engine

[Antoine Pitrou]

Changes by Antoine Pitrou :


--
nosy: +pitrou

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue7593] Computed-goto patch for RE engine

[Georg Brandl]

Changes by Georg Brandl :


--
title: Computed-goto patch -> Computed-goto patch for RE engine

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com