[x265] [PATCH Review only] asm: pixel_add_pp routine for 24x32 block size

[murugan at multicorewareinc.com]

# HG changeset patch
# User Murugan Vairavel <murugan at multicorewareinc.com>
# Date 1383722288 -19800
#      Wed Nov 06 12:48:08 2013 +0530
# Node ID 8678998ac66c1e1bfe2f40ac163a82609fcf14c5
# Parent  4f1d081e7317608eda7db8ce86f72c40568217f9
asm: pixel_add_pp routine for 24x32 block size

diff -r 4f1d081e7317 -r 8678998ac66c source/common/x86/pixel-add8.asm
--- a/source/common/x86/pixel-add8.asm	Wed Nov 06 12:02:03 2013 +0530
+++ b/source/common/x86/pixel-add8.asm	Wed Nov 06 12:48:08 2013 +0530
@@ -393,3 +393,48 @@
 PIXELADD_PP_W16_H4 16, 32
 PIXELADD_PP_W16_H4 16, 64
 
+;-----------------------------------------------------------------------------
+; void pixel_add_pp_c_%1x%2(pixel *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
+;-----------------------------------------------------------------------------
+%macro PIXELADD_PP_W24_H2 2
+INIT_XMM sse2
+cglobal pixel_add_pp_%1x%2, 4, 7, 8, dest, deststride, src0, src1
+
+mov    r4d,    r4m
+mov    r5d,    r5m
+mov    r6d,    %2
+
+.loop
+
+    movu       m0,    [r2]
+    movh       m1,    [r2 + 16]
+    movu       m2,    [r3]
+    movh       m3,    [r3 + 16]
+
+    movu       m4,    [r2 + r4]
+    movh       m5,    [r2 + r4 + 16]
+    movu       m6,    [r3 + r5]
+    movh       m7,    [r3 + r5 + 16]
+
+    paddusb    m0,    m2
+    paddusb    m1,    m3
+    paddusb    m4,    m6
+    paddusb    m5,    m7
+
+    movu    [r0],              m0
+    movh    [r0 + 16],         m1
+    movu    [r0 + r1],         m4
+    movh    [r0 + r1 + 16],    m5
+
+    lea     r2,                [r2 + 2 * r4]
+    lea     r3,                [r3 + 2 * r5]
+    lea     r0,                [r0 + 2 * r1]
+
+    sub     r6d,               2
+
+jnz    .loop
+
+RET
+%endmacro
+
+PIXELADD_PP_W24_H2 24, 32