[PATCH i386]: Enable push/pop in pro/epilogue for modern CPUs

[H.J. Lu]

On Wed, Dec 12, 2012 at 10:21 PM, Jakub Jelinek <jakub@redhat.com> wrote:
> On Wed, Dec 12, 2012 at 10:09:14PM -0800, Xinliang David Li wrote:
>> On Wed, Dec 12, 2012 at 5:19 PM, Jan Hubicka <hubicka@ucw.cz> wrote:
>> >> > libcall is not faster up to 8KB to rep sequence that is better for regalloc/code
>> >> > cache than fully blowin function call.
>> >>
>> >> Be careful with this. My recollection is that REP sequence is good for
>> >> any size -- for smaller size, the REP initial set up cost is too high
>> >> (10s of cycles), while for large size copy, it is less efficient
>> >> compared with library version.
>> >
>> > Well this is based on the data from the memtest script.
>> > Core has good REP implementation - it is a win from rather small blocks (16
>> > bytes if I recall) and it does not need alignment.
>> > Library version starts to be interesting with caching hints, but I think till 80KB
>> > it is still not a win for my setup (glibc-2.15)
>>
>> A simple test shows that -mstringop-strategy=libcall always beats
>> -mstringop-strategy=rep_8byte (on core2 and corei7) except for size
>> smaller than 8 where the rep_8byte strategy simply bypasses REP movs.
>> Can you share your memtest ?
>
> I can't believe that say 16 byte or 32 byte memcpy can be ever faster using a
> libcall.  The PLT call overhead is simply too high.
>

The x86 string/memory functions in the current glibc are
extremely fast and tuned for Core 2/Core i7.  GCC is having
a very hard time to beat them with inlining:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43052

-- 
H.J.