Subject: [PATCH] Add x86-64 memset with unaligned store and rep stosb
H.J. Lu
hjl.tools@gmail.com
Thu Mar 31 16:04:00 GMT 2016
On Wed, Mar 30, 2016 at 4:18 PM, Carlos O'Donell <carlos@redhat.com> wrote:
> On 03/29/2016 03:23 PM, H.J. Lu wrote:
>> The goal of this patch is to replace SSE2 and AVX2 memset.S
>> with faster and smaller alternatives, also support 64-byte vector
>> register size. bench-memset data on various Intel and AMD
>> processors is at
>>
>> https://sourceware.org/bugzilla/show_bug.cgi?id=19881
>>
>> Any comments, feedbacks?
>
> Caveats about Penryn being slower apply here, and I expect your answer
> is the same: the selection of the ifunc will not change, and so Penryn
> will not use the newer versions.
>
> This looks good to me.
>
> Again, same question about thresholding below.
>
Here is the updated patch I am going to check in.
Thanks.
--
H.J.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Add-x86-64-memset-with-unaligned-store-and-rep-stosb.patch
Type: text/x-patch
Size: 14650 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/libc-alpha/attachments/20160331/66bc8dfa/attachment.bin>
More information about the Libc-alpha
mailing list