[x265] [PATCH Review only] asm: pixel_add_pp routine for 2x8 block size

[murugan at multicorewareinc.com]

# HG changeset patch
# User Murugan Vairavel <murugan at multicorewareinc.com>
# Date 1383734569 -19800
#      Wed Nov 06 16:12:49 2013 +0530
# Node ID 2f2d93d4748ce039a24ba3fae412d79769960145
# Parent  377f4e2848d6fa3143154f0b7e7f35bff93cb6a8
asm: pixel_add_pp routine for 2x8 block size

diff -r 377f4e2848d6 -r 2f2d93d4748c source/common/x86/pixel-add8.asm
--- a/source/common/x86/pixel-add8.asm	Wed Nov 06 15:49:19 2013 +0530
+++ b/source/common/x86/pixel-add8.asm	Wed Nov 06 16:12:49 2013 +0530
@@ -71,6 +71,86 @@
 RET
 
 ;-----------------------------------------------------------------------------
+; void pixel_add_pp_c_2x8(pixel *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
+;-----------------------------------------------------------------------------
+INIT_XMM sse2
+cglobal pixel_add_pp_2x8, 4, 7, 8, dest, deststride, src0, src1
+
+mov        r4d,        r4m
+mov        r5d,        r5m
+
+movd       m0,    [r2]
+movd       m1,    [r3]
+
+movd       m2,    [r2 + r4]
+movd       m3,    [r3 + r5]
+
+movd       m4,    [r2 + 2 * r4]
+movd       m5,    [r3 + 2 * r5]
+
+lea        r2,    [r2 + 2 * r4]
+lea        r3,    [r3 + 2 * r5]
+
+movd       m6,    [r2 +  r4]
+movd       m7,    [r3 +  r5]
+
+paddusb    m0,    m1
+paddusb    m2,    m3
+paddusb    m4,    m5
+paddusb    m6,    m7
+
+movd       m1,    [r2 + 2 * r4]
+movd       m3,    [r3 + 2 * r5]
+
+paddusb    m1,    m3
+
+movd    r6d,              m0
+mov     [r0],             r6w
+movd    r6d,              m2
+mov     [r0 + r1],        r6w
+movd    r6d,              m4
+mov     [r0 + 2 * r1],    r6w
+
+lea     r0,               [r0 + 2 * r1]
+
+movd    r6d,              m6
+mov     [r0 + r1],        r6w
+movd    r6d,              m1
+mov     [r0 + 2 * r1],    r6w
+
+lea     r2,               [r2 + 2 * r4]
+lea     r3,               [r3 + 2 * r5]
+lea     r0,               [r0 + 2 * r1]
+
+movd       m0,    [r2  +r4]
+movd       m1,    [r3 + r5]
+
+movd       m2,    [r2 + 2 * r4]
+movd       m3,    [r3 + 2 * r5]
+
+lea        r2,    [r2 + 2 * r4]
+lea        r3,    [r3 + 2 * r5]
+
+movd       m4,    [r2  +r4]
+movd       m5,    [r3 + r5]
+
+paddusb    m0,    m1
+paddusb    m2,    m3
+paddusb    m4,    m5
+
+movd    r6d,              m0
+mov     [r0 + r1],        r6w
+movd    r6d,              m2
+mov     [r0 + 2 * r1],    r6w
+
+lea     r0,               [r0 + 2 * r1]
+
+movd    r6d,              m4
+mov     [r0 + r1],        r6w
+
+RET
+
+;-----------------------------------------------------------------------------
 ; void pixel_add_pp_c_4x2(pixel *dest, intptr_t destride, pixel *src0, pixel *src1, intptr_t srcstride0, intptr_t srcstride1);
 ;-----------------------------------------------------------------------------
 INIT_XMM sse2