[x265] [PATCH] Luma_hpp[64x64 , 64x48 , 64x32 , 64x16] avx2 asm code : improved 33137c->24179 , 24826c->17629 , 16726c->11766 , 7830c->5820c

[aasaipriya at multicorewareinc.com]

# HG changeset patch
# User Aasaipriya Chandran <aasaipriya at multicorewareinc.com>
# Date 1416223817 -19800
#      Mon Nov 17 17:00:17 2014 +0530
# Node ID d225eeaecba038fe09753c6e9a7081a8611b38a8
# Parent  77ff0a5f11d15369ce1c624a507dc78e25cd43fb
Luma_hpp[64x64 , 64x48 , 64x32 , 64x16] avx2 asm code : improved 33137c->24179 , 24826c->17629 , 16726c->11766 , 7830c->5820c

diff -r 77ff0a5f11d1 -r d225eeaecba0 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp	Fri Nov 14 16:20:05 2014 +0530
+++ b/source/common/x86/asm-primitives.cpp	Mon Nov 17 17:00:17 2014 +0530
@@ -1809,6 +1809,11 @@
 		p.luma_hpp[LUMA_32x16] = x265_interp_8tap_horiz_pp_32x16_avx2;
 		p.luma_hpp[LUMA_32x64] = x265_interp_8tap_horiz_pp_32x64_avx2;
 
+	p.luma_hpp[LUMA_64x64] = x265_interp_8tap_horiz_pp_64x64_avx2;
+	p.luma_hpp[LUMA_64x32] = x265_interp_8tap_horiz_pp_64x32_avx2;
+	p.luma_hpp[LUMA_64x48] = x265_interp_8tap_horiz_pp_64x48_avx2;
+	p.luma_hpp[LUMA_64x16] = x265_interp_8tap_horiz_pp_64x16_avx2;
+
 
         p.luma_hpp[LUMA_64x64] = x265_interp_8tap_horiz_pp_64x64_avx2;
         p.luma_vpp[LUMA_4x4] = x265_interp_8tap_vert_pp_4x4_avx2;
diff -r 77ff0a5f11d1 -r d225eeaecba0 source/common/x86/ipfilter8.asm
--- a/source/common/x86/ipfilter8.asm	Fri Nov 14 16:20:05 2014 +0530
+++ b/source/common/x86/ipfilter8.asm	Mon Nov 17 17:00:17 2014 +0530
@@ -1035,6 +1035,117 @@
     RET	
 
 
+    %macro IPFILTER_LUMA_64x_avx2 2
+    INIT_YMM avx2
+cglobal interp_8tap_horiz_pp_%1x%2, 4,6,8
+    sub             r0, 3
+    mov             r4d, r4m
+%ifdef PIC
+    lea             r5, [tab_LumaCoeff]
+    vpbroadcastd    m0, [r5 + r4 * 8]
+    vpbroadcastd    m1, [r5 + r4 * 8 + 4]
+%else
+    vpbroadcastd    m0, [tab_LumaCoeff + r4 * 8]
+    vpbroadcastd    m1, [tab_LumaCoeff + r4 * 8 + 4]
+%endif
+    mova            m2, [tab_Tm]
+    movu            m3, [tab_Tm + 16]
+    vpbroadcastd    m7, [pw_1]
+
+    ; register map
+    ; m0 , m1 interpolate coeff
+    ; m2 , m2  shuffle order table
+    ; m7 - pw_1
+
+    mov             r4d,  %2
+.loop:
+    ; Row 0
+    vbroadcasti128  m4, [r0]                        ; [x E D C B A 9 8 7 6 5 4 3 2 1 0]
+    pshufb          m5, m4, m3
+    pshufb          m4, m2
+    pmaddubsw       m4, m0
+    pmaddubsw       m5, m1
+    paddw           m4, m5
+    pmaddwd         m4, m7
+    vbroadcasti128  m5, [r0 + 8]                    
+    pshufb          m6, m5, m3
+    pshufb          m5, m2
+    pmaddubsw       m5, m0
+    pmaddubsw       m6, m1
+    paddw           m5, m6
+    pmaddwd         m5, m7
+    packssdw        m4, m5                          ; [17 16 15 14 07 06 05 04 13 12 11 10 03 02 01 00]
+    pmulhrsw        m4, [pw_512]
+    packuswb        m4, m4
+    vpermq          m4, m4, 11011000b
+    pshufd          xm4, xm4, 11011000b
+    movu            [r2],xm4
+    vbroadcasti128  m4, [r0 + 16]               
+    pshufb          m5, m4, m3
+    pshufb          m4, m2
+    pmaddubsw       m4, m0
+    pmaddubsw       m5, m1
+    paddw           m4, m5
+    pmaddwd         m4, m7
+    vbroadcasti128  m5, [r0 + 24]                   
+    pshufb          m6, m5, m3
+    pshufb          m5, m2
+    pmaddubsw       m5, m0
+    pmaddubsw       m6, m1
+    paddw           m5, m6
+    pmaddwd         m5, m7
+    packssdw        m4, m5
+    pmulhrsw        m4, [pw_512]
+    packuswb        m4, m4
+    vpermq          m4, m4, 11011000b
+    pshufd          xm4, xm4, 11011000b
+    movu            [r2 + 16],xm4
+    vbroadcasti128  m4, [r0 + 32]
+    pshufb          m5, m4, m3
+    pshufb          m4, m2
+    pmaddubsw       m4, m0
+    pmaddubsw       m5, m1
+    paddw           m4, m5
+    pmaddwd         m4, m7
+    vbroadcasti128  m5, [r0 + 40]
+    pshufb          m6, m5, m3
+    pshufb          m5, m2
+    pmaddubsw       m5, m0
+    pmaddubsw       m6, m1
+    paddw           m5, m6
+    pmaddwd         m5, m7
+    packssdw        m4, m5 
+    pmulhrsw        m4, [pw_512]
+    packuswb        m4, m4
+    vpermq          m4, m4, 11011000b
+    pshufd          xm4, xm4, 11011000b
+    movu            [r2 + 32],xm4
+    vbroadcasti128  m4, [r0 + 48]
+    pshufb          m5, m4, m3
+    pshufb          m4, m2
+    pmaddubsw       m4, m0
+    pmaddubsw       m5, m1
+    paddw           m4, m5
+    pmaddwd         m4, m7
+    vbroadcasti128  m5, [r0 + 56]
+    pshufb          m6, m5, m3
+    pshufb          m5, m2
+    pmaddubsw       m5, m0
+    pmaddubsw       m6, m1
+    paddw           m5, m6
+    pmaddwd         m5, m7
+    packssdw        m4, m5
+    pmulhrsw        m4, [pw_512]
+    packuswb        m4, m4
+    vpermq          m4, m4, 11011000b
+    pshufd          xm4, xm4, 11011000b
+    movu            [r2 + 48],xm4 
+    lea             r0, [r0 + r1]
+    lea             r2, [r2 + r3]
+    dec             r4d
+    jnz             .loop
+    RET
+    %endmacro
 
 
 ;--------------------------------------------------------------------------------------------------------------
@@ -1055,6 +1166,11 @@
     IPFILTER_LUMA_32x_avx2 32 , 16
     IPFILTER_LUMA_32x_avx2 32 , 64
 
+	IPFILTER_LUMA_64x_avx2 64 , 64
+	IPFILTER_LUMA_64x_avx2 64 , 48
+	IPFILTER_LUMA_64x_avx2 64 , 32
+	IPFILTER_LUMA_64x_avx2 64 , 16
+
 ;--------------------------------------------------------------------------------------------------------------
 ; void interp_8tap_horiz_pp_%1x%2(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
 ;--------------------------------------------------------------------------------------------------------------