vq_arm4.h

00001 /* Copyright (C) 2004 Jean-Marc Valin */
00002 /**
00003    @file vq_arm4.h
00004    @brief ARM4-optimized vq routine
00005 */
00006 /*
00007    Redistribution and use in source and binary forms, with or without
00008    modification, are permitted provided that the following conditions
00009    are met:
00010    
00011    - Redistributions of source code must retain the above copyright
00012    notice, this list of conditions and the following disclaimer.
00013    
00014    - Redistributions in binary form must reproduce the above copyright
00015    notice, this list of conditions and the following disclaimer in the
00016    documentation and/or other materials provided with the distribution.
00017    
00018    - Neither the name of the Xiph.org Foundation nor the names of its
00019    contributors may be used to endorse or promote products derived from
00020    this software without specific prior written permission.
00021    
00022    THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
00023    ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
00024    LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
00025    A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR
00026    CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
00027    EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
00028    PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
00029    PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
00030    LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
00031    NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
00032    SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
00033 */
00034 
00035 #define OVERRIDE_VQ_NBEST
00036 void vq_nbest(spx_word16_t *in, const spx_word16_t *codebook, int len, int entries, spx_word32_t *E, int N, int *nbest, spx_word32_t *best_dist, char *stack)
00037 {
00038    int i,j;
00039    for (i=0;i<entries;i+=4)
00040    {
00041 #if 1
00042       spx_word32_t dist1, dist2, dist3, dist4;
00043       int dead1, dead2, dead3, dead4, dead5, dead6, dead7, dead8;
00044       __asm__ __volatile__ (
00045             "mov %0, #0 \n\t"
00046             "mov %1, #0 \n\t"
00047             "mov %2, #0 \n\t"
00048             "mov %3, #0 \n\t"
00049             "mov %10, %4 \n\t"
00050             "add %4, %4, %4\n\t"
00051             ".vqloop%=:\n\t"
00052             "ldrsh %7, [%5], #2 \n\t"
00053             "ldrsh %8, [%6] \n\t"
00054             "mov %9, %6 \n\t"
00055             "mla %0, %7, %8, %0 \n\t"
00056             "ldrsh %8, [%9, %4]! \n\t"
00057             "mla %1, %7, %8, %1 \n\t"
00058             "ldrsh %8, [%9, %4]!\n\t"
00059             "mla %2, %7, %8, %2 \n\t"
00060             "ldrsh %8, [%9, %4]! \n\t"
00061             "mla %3, %7, %8, %3 \n\t"
00062             "subs %10, %10, #1 \n\t"
00063             "add %6, %6, #2 \n\t"
00064             "bne .vqloop%="
00065          : "=r" (dist1), "=r" (dist2), "=r" (dist3), "=r" (dist4),
00066       "=r" (dead1), "=r" (dead2), "=r" (codebook), "=r" (dead4),
00067       "=r" (dead5), "=r" (dead6), "=r" (dead7)
00068          : "4" (len), "5" (in), "6" (codebook)
00069          : "cc");
00070 #else
00071 dist1=dist2=dist3=dist4=0;
00072    /*   spx_word32_t dist1=0;
00073       spx_word32_t dist2=0;
00074       spx_word32_t dist3=0;
00075       spx_word32_t dist4=0;*/
00076       for (j=0;j<2;j++)
00077       {
00078          const spx_word16_t *code = codebook;
00079          dist1 = MAC16_16(dist1,in[j],*code);
00080          code += len;
00081          dist2 = MAC16_16(dist2,in[j],*code);
00082          code += len;
00083          dist3 = MAC16_16(dist3,in[j],*code);
00084          code += len;
00085          dist4 = MAC16_16(dist4,in[j],*code);
00086          codebook++;
00087       }
00088 #endif
00089       dist1=SUB32(SHR(*E++,1),dist1);
00090       if (dist1<*best_dist || i==0)
00091       {
00092          *best_dist=dist1;
00093          *nbest=i;
00094       }
00095       dist2=SUB32(SHR(*E++,1),dist2);
00096       if (dist2<*best_dist)
00097       {
00098          *best_dist=dist2;
00099          *nbest=i+1;
00100       }
00101       dist3=SUB32(SHR(*E++,1),dist3);
00102       if (dist3<*best_dist)
00103       {
00104          *best_dist=dist3;
00105          *nbest=i+2;
00106       }
00107       dist4=SUB32(SHR(*E++,1),dist4);
00108       if (dist4<*best_dist)
00109       {
00110          *best_dist=dist4;
00111          *nbest=i+3;
00112       }
00113       codebook += 3*len;
00114    }
00115 }