[x265] [PATCH 11 of 18] asm: intra_pred_ang4_11_sse2 16-bit

[dtyx265 at gmail.com]

# HG changeset patch
# User David T Yuen <dtyx265 at gmail.com>
# Date 1428076422 25200
# Node ID dfcd48ac5d09a2a0615d5657328457b08ce12e0e
# Parent  89cf1dc260ad91eba0f4edcacc784be0959636b5
asm: intra_pred_ang4_11_sse2 16-bit

This is backported from sse4 code and replaces c code.

./test/TestBench --testbench intrapred | grep "intra_ang_4x4\[11\]"
intra_ang_4x4[11]	4.19x 	 462.50   	 1938.83

transposed mode

./test/TestBench --testbench intrapred | grep "intra_ang_4x4\[25\]"
intra_ang_4x4[25]	3.80x 	 384.97   	 1462.87

diff -r 89cf1dc260ad -r dfcd48ac5d09 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp	Fri Apr 03 08:50:10 2015 -0700
+++ b/source/common/x86/asm-primitives.cpp	Fri Apr 03 08:53:42 2015 -0700
@@ -891,6 +891,8 @@
         p.cu[BLOCK_4x4].intra_pred[8] = x265_intra_pred_ang4_8_sse2;
         p.cu[BLOCK_4x4].intra_pred[9] = x265_intra_pred_ang4_9_sse2;
         p.cu[BLOCK_4x4].intra_pred[10] = x265_intra_pred_ang4_10_sse2;
+        p.cu[BLOCK_4x4].intra_pred[11] = x265_intra_pred_ang4_11_sse2;
+        p.cu[BLOCK_4x4].intra_pred[25] = x265_intra_pred_ang4_11_sse2;
         p.cu[BLOCK_4x4].intra_pred[26] = x265_intra_pred_ang4_26_sse2;
         p.cu[BLOCK_4x4].intra_pred[27] = x265_intra_pred_ang4_9_sse2;
         p.cu[BLOCK_4x4].intra_pred[28] = x265_intra_pred_ang4_8_sse2;
diff -r 89cf1dc260ad -r dfcd48ac5d09 source/common/x86/intrapred16.asm
--- a/source/common/x86/intrapred16.asm	Fri Apr 03 08:50:10 2015 -0700
+++ b/source/common/x86/intrapred16.asm	Fri Apr 03 08:53:42 2015 -0700
@@ -982,6 +982,28 @@
 .quit:
     RET
 
+cglobal intra_pred_ang4_11, 3,5,8
+    xor         r4d, r4d
+    cmp         r3m, byte 25
+    mov         r3d, 16
+    cmove       r3d, r4d
+
+    movh        m1, [r2 + r3 + 2]   ; [x x x 4 3 2 1 0]
+    movh        m2, [r2 - 6]
+    punpcklqdq  m2, m1
+    psrldq      m2, 6
+    punpcklwd   m2, m1          ; [4 3 3 2 2 1 1 0]
+    mova        m3, m2
+    mova        m4, m2
+    mova        m5, m2
+
+    lea         r3, [ang_table + 24 * 16]
+    mova        m0, [r3 +  6 * 16]  ; [24]
+    mova        m1, [r3 +  4 * 16]  ; [26]
+    mova        m6, [r3 +  2 * 16]  ; [28]
+    mova        m7, [r3 +  0 * 16]  ; [30]
+    jmp         mangle(private_prefix %+ _ %+ intra_pred_ang4_3 %+ SUFFIX %+ .do_filter4x4)
+
 ;-----------------------------------------------------------------------------------
 ; void intra_pred_dc(pixel* dst, intptr_t dstStride, pixel* above, int, int filter)
 ;-----------------------------------------------------------------------------------