[x265] [PATCH 2 of 2] asm: intra_pred_ang8_16 improved by ~3% over SSE4

[praveen at multicorewareinc.com]

# HG changeset patch
# User Praveen Tiwari <praveen at multicorewareinc.com>
# Date 1428649728 -19800
#      Fri Apr 10 12:38:48 2015 +0530
# Node ID f76093f3b51852b369aba4d3c738146f869c3151
# Parent  c25751eb5a1ab4b19a171e9f35f149b408c4c19e
asm: intra_pred_ang8_16 improved by ~3% over SSE4

AVX2:
intra_ang_8x8[16]       9.22x    360.04          3320.64

SSE4:
intra_ang_8x8[16]       8.68x    371.05          3222.21

diff -r c25751eb5a1a -r f76093f3b518 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp	Fri Apr 10 12:01:19 2015 +0530
+++ b/source/common/x86/asm-primitives.cpp	Fri Apr 10 12:38:48 2015 +0530
@@ -1775,6 +1775,7 @@
         p.cu[BLOCK_8x8].intra_pred[23] = x265_intra_pred_ang8_23_avx2;
         p.cu[BLOCK_8x8].intra_pred[14] = x265_intra_pred_ang8_14_avx2;
         p.cu[BLOCK_8x8].intra_pred[15] = x265_intra_pred_ang8_15_avx2;
+        p.cu[BLOCK_8x8].intra_pred[16] = x265_intra_pred_ang8_16_avx2;
         p.cu[BLOCK_16x16].intra_pred[3] = x265_intra_pred_ang16_3_avx2;
         p.cu[BLOCK_16x16].intra_pred[4] = x265_intra_pred_ang16_4_avx2;
         p.cu[BLOCK_16x16].intra_pred[5] = x265_intra_pred_ang16_5_avx2;
diff -r c25751eb5a1a -r f76093f3b518 source/common/x86/intrapred.h
--- a/source/common/x86/intrapred.h	Fri Apr 10 12:01:19 2015 +0530
+++ b/source/common/x86/intrapred.h	Fri Apr 10 12:38:48 2015 +0530
@@ -236,6 +236,7 @@
 void x265_intra_pred_ang8_13_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
 void x265_intra_pred_ang8_14_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
 void x265_intra_pred_ang8_15_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
+void x265_intra_pred_ang8_16_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
 void x265_intra_pred_ang8_20_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
 void x265_intra_pred_ang8_21_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
 void x265_intra_pred_ang8_22_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
diff -r c25751eb5a1a -r f76093f3b518 source/common/x86/intrapred8.asm
--- a/source/common/x86/intrapred8.asm	Fri Apr 10 12:01:19 2015 +0530
+++ b/source/common/x86/intrapred8.asm	Fri Apr 10 12:38:48 2015 +0530
@@ -11951,6 +11951,70 @@
     movhps            [r0 + r3], xm2
     RET
 
+INIT_YMM avx2
+cglobal intra_pred_ang8_16, 3, 6, 6
+    mova              m3, [pw_1024]
+    movu              xm5, [r2 + 16]
+    pinsrb            xm5, [r2], 0
+    lea               r5, [intra_pred_shuff_0_8]
+    mova              xm0, xm5
+    pslldq            xm5, 1
+    pinsrb            xm5, [r2 + 2], 0
+    vinserti128       m0, m0, xm5, 1
+    pshufb            m0, [r5]
+
+    lea               r4, [c_ang8_mode_20]
+    pmaddubsw         m1, m0, [r4]
+    pmulhrsw          m1, m3
+    mova              xm0, xm5
+    pslldq            xm5, 1
+    pinsrb            xm5, [r2 + 3], 0
+    vinserti128       m0, m0, xm5, 1
+    pshufb            m0, [r5]
+    pmaddubsw         m2, m0, [r4 + mmsize]
+    pmulhrsw          m2, m3
+    pslldq            xm5, 1
+    pinsrb            xm5, [r2 + 5], 0
+    vinserti128       m0, m5, xm5, 1
+    pshufb            m0, [r5]
+    pmaddubsw         m4, m0, [r4 + 2 * mmsize]
+    pmulhrsw          m4, m3
+    pslldq            xm5, 1
+    pinsrb            xm5, [r2 + 6], 0
+    mova              xm0, xm5
+    pslldq            xm5, 1
+    pinsrb            xm5, [r2 + 8], 0
+    vinserti128       m0, m0, xm5, 1
+    pshufb            m0, [r5]
+    pmaddubsw         m0, [r4 + 3 * mmsize]
+    pmulhrsw          m0, m3
+
+    packuswb          m1, m2
+    packuswb          m4, m0
+
+    vperm2i128        m2, m1, m4, 00100000b
+    vperm2i128        m1, m1, m4, 00110001b
+    punpcklbw         m4, m2, m1
+    punpckhbw         m2, m1
+    punpcklwd         m1, m4, m2
+    punpckhwd         m4, m2
+    mova              m0, [trans8_shuf]
+    vpermd            m1, m0, m1
+    vpermd            m4, m0, m4
+
+    lea               r3, [3 * r1]
+    movq              [r0], xm1
+    movhps            [r0 + r1], xm1
+    vextracti128      xm2, m1, 1
+    movq              [r0 + 2 * r1], xm2
+    movhps            [r0 + r3], xm2
+    lea               r0, [r0 + 4 * r1]
+    movq              [r0], xm4
+    movhps            [r0 + r1], xm4
+    vextracti128      xm2, m4, 1
+    movq              [r0 + 2 * r1], xm2
+    movhps            [r0 + r3], xm2
+    RET
 
 INIT_YMM avx2
 cglobal intra_pred_ang8_20, 3, 6, 6