[x265] [PATCH Review Only] asm code for interp_4tap_vert_pp_2x4

[praveen at multicorewareinc.com]

# HG changeset patch
# User Praveen Tiwari
# Date 1382961770 -19800
# Node ID d62d12b69145ffdfb5d645ebe1f452c294fa5876
# Parent  14ae0b70567b0d78738f2cb064b25c46e437d950
asm code for interp_4tap_vert_pp_2x4

diff -r 14ae0b70567b -r d62d12b69145 source/common/x86/ipfilter8.asm
--- a/source/common/x86/ipfilter8.asm	Mon Oct 28 16:47:51 2013 +0530
+++ b/source/common/x86/ipfilter8.asm	Mon Oct 28 17:32:50 2013 +0530
@@ -38,6 +38,8 @@
 tab_Vm:    db 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1
            db 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3
 
+tab_Cm:    db 0, 2, 1, 3, 0, 2, 1, 3, 0, 2, 1, 3, 0, 2, 1, 3
+
 tab_c_512:  times 8 dw 512
 
 tab_ChromaCoeff: db  0, 64,  0,  0
@@ -723,3 +725,79 @@
 %endmacro
 
 FILTER_V4_W12_H2 12, 16
+
+;-----------------------------------------------------------------------------
+;void interp_4tap_vert_pp_2x4(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
+;-----------------------------------------------------------------------------
+INIT_XMM sse4
+cglobal interp_4tap_vert_pp_2x4, 4, 7, 8
+
+mov         r4d,       r4m
+sub         r0,        r1
+
+%ifdef PIC
+lea         r5,        [tab_ChromaCoeff]
+movd        m0,        [r5 + r4 * 4]
+%else
+movd        m0,        [tab_ChromaCoeff + r4 * 4]
+%endif
+
+pshufb      m0,        [tab_Cm]
+
+mova        m1,        [tab_c_512]
+
+movd        m2,        [r0]
+movd        m3,        [r0 + r1]
+movd        m4,        [r0 + 2 * r1]
+lea         r5,        [r0 + 2 * r1]
+movd        m5,        [r5 + r1]
+
+punpcklbw   m2,        m3
+punpcklbw   m6,        m4,        m5
+punpcklbw   m2,        m6
+
+pmaddubsw   m2,        m0
+
+movd        m6,        [r0 + 4 * r1]
+
+punpcklbw   m3,        m4
+punpcklbw   m7,        m5,        m6
+punpcklbw   m3,        m7
+
+pmaddubsw   m3,        m0
+
+phaddw      m2,        m3
+
+pmulhrsw    m2,        m1
+packuswb    m2,        m2
+
+pextrw      [r2],      m2,  0
+pextrw      [r2 + r3], m2,  2
+
+lea         r5,        [r0 + 4 * r1]
+movd        m2,        [r5 + r1]
+
+punpcklbw   m4,        m5
+punpcklbw   m3,        m6,        m2
+punpcklbw   m4,        m3
+
+pmaddubsw   m4,        m0
+
+movd        m3,        [r5 + 2 * r1]
+
+punpcklbw   m5,        m6
+punpcklbw   m2,        m3
+punpcklbw   m5,        m2
+
+pmaddubsw   m5,        m0
+
+phaddw      m4,        m5
+
+pmulhrsw    m4,        m1
+packuswb    m4,        m4
+
+pextrw      [r2 + 2 * r3],    m4,    0
+lea         r6,               [r2 + 2 * r3]
+pextrw      [r6 + r3],        m4,    2
+
+RET