Update to opus-HEAD-66611f1.

silk/x86/VQ_WMat_EC_sse.c

Index: silk/x86/VQ_WMat_EC_sse.c
diff --git a/silk/VQ_WMat_EC.c b/silk/x86/VQ_WMat_EC_sse.c
similarity index 55%
copy from silk/VQ_WMat_EC.c
copy to silk/x86/VQ_WMat_EC_sse.c
index 13d5d34eddbb5504407635af7b865c57c487342e..74d6c6d0ec6e4e9a7414cd4de87de9727ac20688 100644
--- a/silk/VQ_WMat_EC.c
+++ b/silk/x86/VQ_WMat_EC_sse.c
@@ -1,38 +1,42 @@
-/***********************************************************************
-Copyright (c) 2006-2011, Skype Limited. All rights reserved.
-Redistribution and use in source and binary forms, with or without
-modification, are permitted provided that the following conditions
-are met:
-- Redistributions of source code must retain the above copyright notice,
-this list of conditions and the following disclaimer.
-- Redistributions in binary form must reproduce the above copyright
-notice, this list of conditions and the following disclaimer in the
-documentation and/or other materials provided with the distribution.
-- Neither the name of Internet Society, IETF or IETF Trust, nor the
-names of specific contributors, may be used to endorse or promote
-products derived from this software without specific prior written
-permission.
-THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
-AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
-IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
-ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
-LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
-CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
-SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
-INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
-CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
-ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
-POSSIBILITY OF SUCH DAMAGE.
-***********************************************************************/
+/* Copyright (c) 2014, Cisco Systems, INC
+   Written by XiangMingZhu WeiZhou MinPeng YanWang
+
+   Redistribution and use in source and binary forms, with or without
+   modification, are permitted provided that the following conditions
+   are met:
+
+   - Redistributions of source code must retain the above copyright
+   notice, this list of conditions and the following disclaimer.
+
+   - Redistributions in binary form must reproduce the above copyright
+   notice, this list of conditions and the following disclaimer in the
+   documentation and/or other materials provided with the distribution.
+
+   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+   ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
+   OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+   EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+   PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
 
 #ifdef HAVE_CONFIG_H
 #include "config.h"
 #endif
 
+#include <xmmintrin.h>
+#include <emmintrin.h>
+#include <smmintrin.h>
 #include "main.h"
+#include "celt/x86/x86cpu.h"
 
 /* Entropy constrained matrix-weighted VQ, hard-coded to 5-element vectors, for a single input data vector */
-void silk_VQ_WMat_EC(
+void silk_VQ_WMat_EC_sse4_1(
     opus_int8                   *ind,                           /* O    index of best codebook vector               */
     opus_int32                  *rate_dist_Q14,                 /* O    best weighted quant error + mu * rate       */
     opus_int                    *gain_Q7,                       /* O    sum of absolute LTP coefficients            */
@@ -51,32 +55,50 @@ void silk_VQ_WMat_EC(
     opus_int16 diff_Q14[ 5 ];
     opus_int32 sum1_Q14, sum2_Q16;
 
+    __m128i C_tmp1, C_tmp2, C_tmp3, C_tmp4, C_tmp5;
     /* Loop over codebook */
     *rate_dist_Q14 = silk_int32_MAX;
     cb_row_Q7 = cb_Q7;
     for( k = 0; k < L; k++ ) {
-	    gain_tmp_Q7 = cb_gain_Q7[k];
+        gain_tmp_Q7 = cb_gain_Q7[k];
 
         diff_Q14[ 0 ] = in_Q14[ 0 ] - silk_LSHIFT( cb_row_Q7[ 0 ], 7 );
-        diff_Q14[ 1 ] = in_Q14[ 1 ] - silk_LSHIFT( cb_row_Q7[ 1 ], 7 );
-        diff_Q14[ 2 ] = in_Q14[ 2 ] - silk_LSHIFT( cb_row_Q7[ 2 ], 7 );
-        diff_Q14[ 3 ] = in_Q14[ 3 ] - silk_LSHIFT( cb_row_Q7[ 3 ], 7 );
-        diff_Q14[ 4 ] = in_Q14[ 4 ] - silk_LSHIFT( cb_row_Q7[ 4 ], 7 );
+
+        C_tmp1 = OP_CVTEPI16_EPI32_M64( &in_Q14[ 1 ] );
+        C_tmp2 = OP_CVTEPI8_EPI32_M32( &cb_row_Q7[ 1 ] );
+        C_tmp2 = _mm_slli_epi32( C_tmp2, 7 );
+        C_tmp1 = _mm_sub_epi32( C_tmp1, C_tmp2 );
+
+        diff_Q14[ 1 ] = _mm_extract_epi16( C_tmp1, 0 );
+        diff_Q14[ 2 ] = _mm_extract_epi16( C_tmp1, 2 );
+        diff_Q14[ 3 ] = _mm_extract_epi16( C_tmp1, 4 );
+        diff_Q14[ 4 ] = _mm_extract_epi16( C_tmp1, 6 );
 
         /* Weighted rate */
         sum1_Q14 = silk_SMULBB( mu_Q9, cl_Q5[ k ] );
 
-		/* Penalty for too large gain */
-		sum1_Q14 = silk_ADD_LSHIFT32( sum1_Q14, silk_max( silk_SUB32( gain_tmp_Q7, max_gain_Q7 ), 0 ), 10 );
+        /* Penalty for too large gain */
+        sum1_Q14 = silk_ADD_LSHIFT32( sum1_Q14, silk_max( silk_SUB32( gain_tmp_Q7, max_gain_Q7 ), 0 ), 10 );
 
         silk_assert( sum1_Q14 >= 0 );
 
         /* first row of W_Q18 */
-        sum2_Q16 = silk_SMULWB(           W_Q18[  1 ], diff_Q14[ 1 ] );
-        sum2_Q16 = silk_SMLAWB( sum2_Q16, W_Q18[  2 ], diff_Q14[ 2 ] );
-        sum2_Q16 = silk_SMLAWB( sum2_Q16, W_Q18[  3 ], diff_Q14[ 3 ] );
-        sum2_Q16 = silk_SMLAWB( sum2_Q16, W_Q18[  4 ], diff_Q14[ 4 ] );
-        sum2_Q16 = silk_LSHIFT( sum2_Q16, 1 );
+        C_tmp3 = _mm_loadu_si128( (__m128i *)(&W_Q18[ 1 ] ) );
+        C_tmp4 = _mm_mul_epi32( C_tmp3, C_tmp1 );
+        C_tmp4 = _mm_srli_si128( C_tmp4, 2 );
+
+        C_tmp1 = _mm_shuffle_epi32( C_tmp1, _MM_SHUFFLE( 0, 3, 2, 1 ) ); /* shift right 4 bytes */
+        C_tmp3 = _mm_shuffle_epi32( C_tmp3, _MM_SHUFFLE( 0, 3, 2, 1 ) ); /* shift right 4 bytes */
+
+        C_tmp5 = _mm_mul_epi32( C_tmp3, C_tmp1 );
+        C_tmp5 = _mm_srli_si128( C_tmp5, 2 );
+
+        C_tmp5 = _mm_add_epi32( C_tmp4, C_tmp5 );
+        C_tmp5 = _mm_slli_epi32( C_tmp5, 1 );
+
+        C_tmp5 = _mm_add_epi32( C_tmp5, _mm_shuffle_epi32( C_tmp5, _MM_SHUFFLE( 0, 0, 0, 2 ) ) );
+        sum2_Q16 = _mm_cvtsi128_si32( C_tmp5 );
+
         sum2_Q16 = silk_SMLAWB( sum2_Q16, W_Q18[  0 ], diff_Q14[ 0 ] );
         sum1_Q14 = silk_SMLAWB( sum1_Q14, sum2_Q16,    diff_Q14[ 0 ] );
 
@@ -111,7 +133,7 @@ void silk_VQ_WMat_EC(
         if( sum1_Q14 < *rate_dist_Q14 ) {
             *rate_dist_Q14 = sum1_Q14;
             *ind = (opus_int8)k;
-			*gain_Q7 = gain_tmp_Q7;
+            *gain_Q7 = gain_tmp_Q7;
         }
 
         /* Go to next cbk vector */