I am seeing no code-size difference on ARM with and without the jump
eliminator.  I'm going to remove it, and replace it with an option to
print unneeded jumps that we see while doing fixups.

Thanks

On Mon, Sep 29, 2008 at 10:08 AM, Dean McNamee <[EMAIL PROTECTED]> wrote:
> I will try to look into ARM, and if it's not an advantage there, and
> ok with Ivan, I will remove ARM jump elimination also.
>
> On Mon, Sep 29, 2008 at 6:31 AM, Kasper Lund <[EMAIL PROTECTED]> wrote:
>> Given the complexity of the code and the very small benefit it gives
>> us, this change certainly LGTM. As you say, we should probably do the
>> same on ARM.
>>
>> On Sun, Sep 28, 2008 at 3:07 PM,  <[EMAIL PROTECTED]> wrote:
>>> Reviewers: Kasper Lund, iposva,
>>>
>>> Message:
>>> I realize I should probably look at ARM also, remove the jump
>>> elimination flag all together, or make it ARM specific.  I just wanted
>>> to wait until I got some initial feedback, maybe you think this is a bad
>>> idea.
>>>
>>> The single jump that was being eliminated with this current code was in
>>> the fast switch code.  I didn't track it down all the way, but it might
>>> be easy to fix statically.  Otherwise, it's not a huge deal, we're going
>>> to save a lot more than 5 bytes with short jumps.
>>>
>>> I could try to keep the code and make it work with short jump encoding.
>>> It just seemed to be a lot of complexity for hardly ever saving a few
>>> bytes at most in my experiments.
>>>
>>> Description:
>>> Remove x86 jump elimination.  This was just complicated enough to make
>>> it annoying to support short jump encoding.  I measured a code size
>>> increase of 5 bytes on the V8 benchmark, from missing one jump to next
>>> elimination possibility.
>>>
>>> Please review this at http://codereview.chromium.org/4321
>>>
>>> Affected files:
>>>  M src/assembler-ia32.cc
>>>  M src/assembler-ia32.h
>>>
>>>
>>>
>>
>

--~--~---------~--~----~------------~-------~--~----~
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
-~----------~----~----~----~------~----~------~--~---

Reply via email to