[x265] [PATCH] asm: assembly code for pixel_sa8d_32x16

[yuvaraj at multicorewareinc.com]

# HG changeset patch
# User Yuvaraj Venkatesh <yuvaraj at multicorewareinc.com>
# Date 1384950142 -19800
#      Wed Nov 20 17:52:22 2013 +0530
# Node ID cba4b1be723eb8d31d4d9499d1324b86e23b26af
# Parent  7b914c772f8a0bda2df22dff541e4671eeb9f80c
asm: assembly code for pixel_sa8d_32x16

diff -r 7b914c772f8a -r cba4b1be723e source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp	Wed Nov 20 17:08:22 2013 +0530
+++ b/source/common/x86/asm-primitives.cpp	Wed Nov 20 17:52:22 2013 +0530
@@ -98,7 +98,7 @@
     p.sa8d[BLOCK_32x32] = x265_pixel_sa8d_32x32_ ## cpu; \
     p.sa8d[BLOCK_64x64] = x265_pixel_sa8d_64x64_ ## cpu; \
     p.sa8d_inter[LUMA_32x32] = x265_pixel_sa8d_32x32_ ## cpu; \
-    p.sa8d_inter[LUMA_32x16] = cmp<32, 16, 16, 16, x265_pixel_sa8d_16x16_ ## cpu>; \
+    p.sa8d_inter[LUMA_32x16] = x265_pixel_sa8d_32x16_ ## cpu; \
     p.sa8d_inter[LUMA_16x32] = cmp<16, 32, 16, 16, x265_pixel_sa8d_16x16_ ## cpu>; \
     p.sa8d_inter[LUMA_64x64] = x265_pixel_sa8d_64x64_ ## cpu; \
     p.sa8d_inter[LUMA_64x32] = cmp<64, 32, 16, 16, x265_pixel_sa8d_16x16_ ## cpu>; \
diff -r 7b914c772f8a -r cba4b1be723e source/common/x86/pixel-a.asm
--- a/source/common/x86/pixel-a.asm	Wed Nov 20 17:08:22 2013 +0530
+++ b/source/common/x86/pixel-a.asm	Wed Nov 20 17:52:22 2013 +0530
@@ -3583,6 +3583,27 @@
     shr  eax, 1
     RET
 
+cglobal pixel_sa8d_32x16, 4,8,12
+    FIX_STRIDES r1, r3
+    lea  r4, [3*r1]
+    lea  r5, [3*r3]
+    pxor m12, m12
+%if vertical == 0
+    mova m7, [hmul_8p]
+%endif
+    SA8D_16x16
+    lea  r4, [8*r1]
+    lea  r5, [8*r3]
+    sub  r2, r4
+    sub  r0, r5
+    add  r2, 16
+    add  r0, 16
+    lea  r4, [3*r1]
+    lea  r5, [3*r3]
+    SA8D_16x16
+    movd eax, m12
+    RET
+
 cglobal pixel_sa8d_32x32, 4,8,12
     FIX_STRIDES r1, r3
     lea  r4, [3*r1]
@@ -3892,6 +3913,64 @@
     mov  esp, r6
     RET
 
+cglobal pixel_sa8d_32x16, 4,7,8
+    FIX_STRIDES r1, r3
+    mov  r6, esp
+    and  esp, ~15
+    sub  esp, 64
+
+    lea  r4, [r1 + 2*r1]
+    lea  r5, [r3 + 2*r3]
+    call pixel_sa8d_8x8_internal2
+    mova [rsp+48], m0
+    call pixel_sa8d_8x8_internal2
+    SA8D_INTER
+    mova [esp+48], m0
+
+    mov  r0, [r6+20]
+    mov  r2, [r6+28]
+    add  r0, 8*SIZEOF_PIXEL
+    add  r2, 8*SIZEOF_PIXEL
+    call pixel_sa8d_8x8_internal2
+    SA8D_INTER
+    mova [esp+48], m0
+    call pixel_sa8d_8x8_internal2
+    paddusw m0, [esp+48]
+    HADDUW m0, m1
+    movd r4d, m0
+    add  r4d, 1
+    shr  r4d, 1
+    mov dword [esp+36], r4d
+
+    mov  r0, [r6+20]
+    mov  r2, [r6+28]
+    add  r0, 16*SIZEOF_PIXEL
+    add  r2, 16*SIZEOF_PIXEL
+    lea  r4, [r1 + 2*r1]
+    call pixel_sa8d_8x8_internal2
+    mova [esp+48], m0
+    call pixel_sa8d_8x8_internal2
+    SA8D_INTER
+    mova [esp+48], m0
+
+    mov  r0, [r6+20]
+    mov  r2, [r6+28]
+    add  r0, 24*SIZEOF_PIXEL
+    add  r2, 24*SIZEOF_PIXEL
+    call pixel_sa8d_8x8_internal2
+    SA8D_INTER
+    mova [esp+64-mmsize], m0
+    call pixel_sa8d_8x8_internal2
+    paddusw m0, [esp+48]
+    HADDUW m0, m1
+    movd r4d, m0
+    add  r4d, 1
+    shr  r4d, 1
+    add r4d, dword [esp+36]
+    mov eax, r4d
+    mov esp, r6
+    RET
+
 cglobal pixel_sa8d_32x32, 4,7,8
     FIX_STRIDES r1, r3
     mov  r6, esp