[x265] [PATCH] replace sad_24 vector class function with intrinsic

[chen]

At 2013-10-03 21:39:05,dnyaneshwar at multicorewareinc.com wrote:
># HG changeset patch
># User Dnyaneshwar
># Date 1380807452 -19800
>#      Thu Oct 03 19:07:32 2013 +0530
># Node ID 3c61b10dfb1370f0c5479b714258094e88583a01
># Parent  55bf7d1fc1716bc67aba7285deb27542796a56fd
>replace sad_24 vector class function with intrinsic.
>
>+
>+        T10 = _mm_load_si128((__m128i*)(fref));
>+        T11 = _mm_load_si128((__m128i*)(fref + frefstride));
>+        T12 = _mm_load_si128((__m128i*)(fref + (2) * frefstride));
>+        T13 = _mm_load_si128((__m128i*)(fref + (3) * frefstride));
I guess alignment read instruction can't be here. 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/x265-devel/attachments/20131005/793fd4a4/attachment.html>