[x265] [PATCH 04 of 11] asm: intra_pred_ang4_12_sse2

[dtyx265 at gmail.com]

# HG changeset patch
# User David T Yuen <dtyx265 at gmail.com>
# Date 1427505936 25200
# Node ID 3ea097e4acb7e6ce9161b79968c3653741003daf
# Parent  4314cdf34f36c7a7fc3a4b3e744cd05c6ab90352
asm: intra_pred_ang4_12_sse2

This is backported from sse4 code and replaces c code.

64-bit

./test/TestBench --testbench intrapred | grep "intra_ang_4x4\[12\]"
intra_ang_4x4[12]	3.35x 	 617.45   	 2070.22

32-bit

./test/TestBench --testbench intrapred | grep "intra_ang_4x4\[12\]"
intra_ang_4x4[12]	3.87x 	 757.54   	 2935.42

diff -r 4314cdf34f36 -r 3ea097e4acb7 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp	Fri Mar 27 18:20:57 2015 -0700
+++ b/source/common/x86/asm-primitives.cpp	Fri Mar 27 18:25:36 2015 -0700
@@ -1206,6 +1206,7 @@
         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[12] = x265_intra_pred_ang4_12_sse2;
         p.cu[BLOCK_4x4].intra_pred[26] = x265_intra_pred_ang4_26_sse2;
 
         p.cu[BLOCK_4x4].calcresidual = x265_getResidual4_sse2;
diff -r 4314cdf34f36 -r 3ea097e4acb7 source/common/x86/intrapred.h
--- a/source/common/x86/intrapred.h	Fri Mar 27 18:20:57 2015 -0700
+++ b/source/common/x86/intrapred.h	Fri Mar 27 18:25:36 2015 -0700
@@ -57,6 +57,7 @@
 DECL_ANG(4, 9, sse2);
 DECL_ANG(4, 10, sse2);
 DECL_ANG(4, 11, sse2);
+DECL_ANG(4, 12, sse2);
 DECL_ANG(4, 26, sse2);
 
 DECL_ANG(4, 2, ssse3);
diff -r 4314cdf34f36 -r 3ea097e4acb7 source/common/x86/intrapred8.asm
--- a/source/common/x86/intrapred8.asm	Fri Mar 27 18:20:57 2015 -0700
+++ b/source/common/x86/intrapred8.asm	Fri Mar 27 18:25:36 2015 -0700
@@ -1574,6 +1574,27 @@
     mova        m7, [r3 +  0 * 16]  ; [30]
     jmp         mangle(private_prefix %+ _ %+ intra_pred_ang4_3 %+ SUFFIX %+ .do_filter4x4)
 
+cglobal intra_pred_ang4_12, 4,5,8
+    xor         r4d, r4d
+    cmp         r3m, byte 24
+    mov         r3d, 8
+    cmove       r3d, r4d
+
+    movd        m1, [r2 + r3 + 1]
+    movh        m0, [r2 - 7]
+    punpcklbw   m1, m1
+    punpcklqdq  m0, m1
+    psrldq      m0, 7
+    punpcklqdq  m0, m0
+    mova        m2, m0
+
+    lea         r3, [pw_ang_table + 20 * 16]
+    mova        m4, [r3 +  7 * 16]  ; [27]
+    mova        m5, [r3 +  2 * 16]  ; [22]
+    mova        m6, [r3 -  3 * 16]  ; [17]
+    mova        m7, [r3 -  8 * 16]  ; [12]
+    jmp         mangle(private_prefix %+ _ %+ intra_pred_ang4_3 %+ SUFFIX %+ .do_filter4x4)
+
 ;---------------------------------------------------------------------------------------------
 ; void intra_pred_dc(pixel* dst, intptr_t dstStride, pixel *srcPix, int dirMode, int bFilter)
 ;---------------------------------------------------------------------------------------------