[x265] [PATCH 9 of 9] asm: intra_pred_ang8_21 improved by ~5% over SSE4

[praveen at multicorewareinc.com]

# HG changeset patch
# User Praveen Tiwari <praveen at multicorewareinc.com>
# Date 1428583713 -19800
#      Thu Apr 09 18:18:33 2015 +0530
# Node ID 759643fade74e82075a9a6491c41d9f3563df7e2
# Parent  72c75090a5dcbe002bd28d2190703b6d74ac7c81
asm: intra_pred_ang8_21 improved by ~5% over SSE4

AVX2:
intra_ang_8x8[21]       8.55x    239.75          2050.08

SSE4:
intra_ang_8x8[21]       8.03x    252.60          2027.91

diff -r 72c75090a5dc -r 759643fade74 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp	Thu Apr 09 18:07:22 2015 +0530
+++ b/source/common/x86/asm-primitives.cpp	Thu Apr 09 18:18:33 2015 +0530
@@ -1765,6 +1765,7 @@
         p.cu[BLOCK_8x8].intra_pred[24] = x265_intra_pred_ang8_24_avx2;
         p.cu[BLOCK_8x8].intra_pred[11] = x265_intra_pred_ang8_11_avx2;
         p.cu[BLOCK_8x8].intra_pred[13] = x265_intra_pred_ang8_13_avx2;
+        p.cu[BLOCK_8x8].intra_pred[21] = x265_intra_pred_ang8_21_avx2;
         p.cu[BLOCK_8x8].intra_pred[22] = x265_intra_pred_ang8_22_avx2;
         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;
diff -r 72c75090a5dc -r 759643fade74 source/common/x86/intrapred.h
--- a/source/common/x86/intrapred.h	Thu Apr 09 18:07:22 2015 +0530
+++ b/source/common/x86/intrapred.h	Thu Apr 09 18:18:33 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_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);
 void x265_intra_pred_ang8_23_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
 void x265_intra_pred_ang16_3_avx2(pixel* dst, intptr_t dstStride, const pixel* srcPix, int dirMode, int bFilter);
diff -r 72c75090a5dc -r 759643fade74 source/common/x86/intrapred8.asm
--- a/source/common/x86/intrapred8.asm	Thu Apr 09 18:07:22 2015 +0530
+++ b/source/common/x86/intrapred8.asm	Thu Apr 09 18:18:33 2015 +0530
@@ -11945,6 +11945,57 @@
     movhps            [r0 + r3], xm2
     RET
 
+INIT_YMM avx2
+cglobal intra_pred_ang8_21, 3, 6, 6
+    mova              m3, [pw_1024]
+    movu              xm5, [r2]
+    lea               r5, [intra_pred_shuff_0_8]
+    mova              xm0, xm5
+    pslldq            xm5, 1
+    pinsrb            xm5, [r2 + 2 + 16], 0
+    vinserti128       m0, m0, xm5, 1
+    pshufb            m0, [r5]
+
+    lea               r4, [c_ang8_mode_15]
+    pmaddubsw         m1, m0, [r4]
+    pmulhrsw          m1, m3
+    mova              xm0, xm5
+    pslldq            xm5, 1
+    pinsrb            xm5, [r2 + 4 + 16], 0
+    vinserti128       m0, m0, xm5, 1
+    pshufb            m0, [r5]
+    pmaddubsw         m2, m0, [r4 + mmsize]
+    pmulhrsw          m2, m3
+    mova              xm0, xm5
+    pslldq            xm5, 1
+    pinsrb            xm5, [r2 + 6 + 16], 0
+    vinserti128       m0, m0, xm5, 1
+    pshufb            m0, [r5]
+    pmaddubsw         m4, m0, [r4 + 2 * mmsize]
+    pmulhrsw          m4, m3
+    mova              xm0, xm5
+    pslldq            xm5, 1
+    pinsrb            xm5, [r2 + 8 + 16], 0
+    vinserti128       m0, m0, xm5, 1
+    pshufb            m0, [r5]
+    pmaddubsw         m0, [r4 + 3 * mmsize]
+    pmulhrsw          m0, m3
+    packuswb          m1, m2
+    packuswb          m4, m0
+
+    lea               r3, [3 * r1]
+    movq              [r0], xm1
+    vextracti128      xm2, m1, 1
+    movq              [r0 + r1], xm2
+    movhps            [r0 + 2 * r1], xm1
+    movhps            [r0 + r3], xm2
+    lea               r0, [r0 + 4 * r1]
+    movq              [r0], xm4
+    vextracti128      xm2, m4, 1
+    movq              [r0 + r1], xm2
+    movhps            [r0 + 2 * r1], xm4
+    movhps            [r0 + r3], xm2
+    RET
 
 INIT_YMM avx2
 cglobal intra_pred_ang8_22, 3, 6, 6