[x265] [PATCH] asm: 10bpp code for transpose 16x16

[murugan at multicorewareinc.com]

# HG changeset patch
# User Murugan Vairavel <murugan at multicorewareinc.com>
# Date 1386138979 -19800
#      Wed Dec 04 12:06:19 2013 +0530
# Node ID 8b73b22d90e1a0d70495e8b5f009a9c4fc37f258
# Parent  55c0bf9d99661073a7acdb5749e2625379d8393a
asm: 10bpp code for transpose 16x16

diff -r 55c0bf9d9966 -r 8b73b22d90e1 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp	Tue Dec 03 14:14:44 2013 -0600
+++ b/source/common/x86/asm-primitives.cpp	Wed Dec 04 12:06:19 2013 +0530
@@ -528,6 +528,7 @@
 
         p.transpose[BLOCK_4x4] = x265_transpose4_sse2;
         p.transpose[BLOCK_8x8] = x265_transpose8_sse2;
+        p.transpose[BLOCK_16x16] = x265_transpose16_sse2;
 
         p.ssim_4x4x2_core = x265_pixel_ssim_4x4x2_core_sse2;
         PIXEL_AVG(sse2);
diff -r 55c0bf9d9966 -r 8b73b22d90e1 source/common/x86/pixel-util8.asm
--- a/source/common/x86/pixel-util8.asm	Tue Dec 03 14:14:44 2013 -0600
+++ b/source/common/x86/pixel-util8.asm	Wed Dec 04 12:06:19 2013 +0530
@@ -883,7 +883,10 @@
     lea    r1,    [r1 + 2 * r2]
     lea    r0,    [r3 + 8]
     TRANSPOSE_4x4 r5
-    lea    r1,    [r4 + 8]
+    lea    r1,    [r1 + 2 * r2]
+    neg    r2
+    lea    r1,    [r1 + r2 * 8 + 8]
+    neg    r2
     lea    r0,    [r3 + 4 * r5]
     TRANSPOSE_4x4 r5
     lea    r1,    [r1 + 2 * r2]
@@ -893,7 +896,6 @@
 cglobal transpose8, 3, 6, 4, dest, src, stride
     add    r2,    r2
     mov    r3,    r0
-    mov    r4,    r1
     mov    r5,    16
     call   transpose8_internal
 %else
@@ -978,8 +980,29 @@
 ; void transpose_16x16(pixel *dst, pixel *src, intptr_t stride)
 ;-----------------------------------------------------------------
 INIT_XMM sse2
+%if HIGH_BIT_DEPTH
+cglobal transpose16, 3, 7, 4, dest, src, stride
+    add    r2,    r2
+    mov    r3,    r0
+    mov    r4,    r1
+    mov    r5,    32
+    mov    r6,    r0
+    call   transpose8_internal
+    lea    r1,    [r1 - 8 + 2 * r2]
+    lea    r0,    [r6 + 16]
+    mov    r3,    r0
+    call   transpose8_internal
+    lea    r1,    [r4 + 16]
+    lea    r0,    [r6 + 8 * r5]
+    mov    r3,    r0
+    call   transpose8_internal
+    lea    r1,    [r1 - 8 + 2 * r2]
+    lea    r0,    [r6 + 8 * r5 + 16]
+    mov    r3,    r0
+    call   transpose8_internal
+
+%else
 cglobal transpose16, 3, 5, 8, dest, src, stride
-
     mov    r3,    r0
     mov    r4,    r1
     TRANSPOSE_8x8 16
@@ -992,7 +1015,7 @@
     lea    r1,    [r1 + 2 * r2]
     lea    r0,    [r3 + 8 * 16 + 8]
     TRANSPOSE_8x8 16
-
+%endif
     RET
 
 cglobal transpose16_internal