[lfs-dev] why gcc-4.9.0 create larger files on 32bits system than 64btis.

xinglp xinglp at gmail.com
Sat Apr 26 22:11:13 PDT 2014


2014-04-27 13:10 GMT+08:00 xinglp <xinglp at gmail.com>:
> 2014-04-27 11:58 GMT+08:00 Bruce Dubbs <bruce.dubbs at gmail.com>:
>> xinglp wrote:
>>>
>>> For example, on my newly built lfs system:
>>> $ ls -lh
>>> {2014-04-27-3.14.1-x86_64,2014-04-27-3.14.1-i686}/bin/{xz,bash,sync}
>>> -rwxr-xr-x 1 root root 806K Apr 26 23:54 2014-04-27-3.14.1-i686/bin/bash
>>> -rwxr-xr-x 1 root root 747K Apr 27 04:25 2014-04-27-3.14.1-x86_64/bin/bash
>>> -rwxr-xr-x 1 root root  22K Apr 26 23:54 2014-04-27-3.14.1-i686/bin/sync
>>> -rwxr-xr-x 1 root root  24K Apr 27 04:25 2014-04-27-3.14.1-x86_64/bin/sync
>>> -rwxr-xr-x 1 root root  67K Apr 26 23:54 2014-04-27-3.14.1-i686/bin/xz
>>> -rwxr-xr-x 1 root root  64K Apr 27 04:25 2014-04-27-3.14.1-x86_64/bin/xz
>>> The "i686" version files are bigger than the "x86_64". Why ?
>>
>>
>> I have no idea why.
>>
>> On an older 686 system I have /bin/bash as 684K (gcc 4.7.0)
>> On x86_64 it is 816K (gcc 4.8.2)
>> and also        859K (gcc 4.9.0)
>>
>> About the only thing I can think of is the possibility of optimization with
>> loop unrolling or some such.
> But why not the gcc of x86_64 do the same optimization. I feel it
> unnormal that i686 is bigger than x86_64.
But the way, I'm using -O2
>>
>>   -- Bruce
>> --
>> http://lists.linuxfromscratch.org/listinfo/lfs-dev
>> FAQ: http://www.linuxfromscratch.org/faq/
>> Unsubscribe: See the above information page


More information about the lfs-dev mailing list