[issue4715] optimize bytecode for conditional branches

[Collin Winter]

Collin Winter <collinw at gmail.com> added the comment:

On Tue, Jan 13, 2009 at 3:25 PM, Antoine Pitrou <report at bugs.python.org> wrote:
>
> Antoine Pitrou <pitrou at free.fr> added the comment:
>
> Hello,
>
>> I've backported condbranches-plus.patch to trunk, and I'm getting these
>> results:
>
> Thanks!
>
>> PyBench: 1.84-2.21% faster
>> 2to3: 3.83% faster
>> Spitfire: 6.13-6.23% faster
>
> What is Spitfire?

http://code.google.com/p/spitfire/. It's a template system designed
for performance that I have some experience with.

>> I've haven't tested condbranches.patch vs condbranches-plus.patch; what
>> difference are you seeing, Antoine?
>
> condbranches.patch is the earlier version (without POP_OR_JUMP and
> JUMP_OR_POP), it can be ignored.

I was mostly curious whether the POP_OR_JUMP and JUMP_OR_POP opcodes
had a noticeable performance impact, ie, do they make things fast
enough to warrant their inclusion over the old JUMP_IF_FALSE
implementations.

>> This patch mostly looks good, though you still need to change Doc/
>> library/dis.rst and the pure-Python compiler package.
>
> Well, the pure-Python compiler package doesn't exist in py3k, for which
> I've initially made the patch.
> (and its state in 2.x is very sorry...)

I'll update the compiler package in 2.x and post my patch once I have
it working. There are also some test failures in 2.x that I'll take
care of.

_______________________________________
Python tracker <report at bugs.python.org>
<http://bugs.python.org/issue4715>
_______________________________________