Main Page | Class List | File List | Class Members | File Members

ltp_sse.h

Go to the documentation of this file.
00001 /* Copyright (C) 2002 Jean-Marc Valin 00002 File: ltp.c 00003 Lont-Term Prediction functions 00004 00005 Redistribution and use in source and binary forms, with or without 00006 modification, are permitted provided that the following conditions 00007 are met: 00008 00009 - Redistributions of source code must retain the above copyright 00010 notice, this list of conditions and the following disclaimer. 00011 00012 - Redistributions in binary form must reproduce the above copyright 00013 notice, this list of conditions and the following disclaimer in the 00014 documentation and/or other materials provided with the distribution. 00015 00016 - Neither the name of the Xiph.org Foundation nor the names of its 00017 contributors may be used to endorse or promote products derived from 00018 this software without specific prior written permission. 00019 00020 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 00021 ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 00022 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 00023 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR 00024 CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, 00025 EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, 00026 PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR 00027 PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF 00028 LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING 00029 NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 00030 SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 00031 */ 00032 00033 00034 static float inner_prod(float *a, float *b, int len) 00035 { 00036 float sum; 00037 __asm__ __volatile__ ( 00038 "\tpush %%eax\n" 00039 "\tpush %%edi\n" 00040 "\tpush %%ecx\n" 00041 "\txorps %%xmm3, %%xmm3\n" 00042 "\txorps %%xmm4, %%xmm4\n" 00043 00044 "\tsub $20, %%ecx\n" 00045 00046 ".mul20_loop%=:\n" 00047 00048 "\tmovups (%%eax), %%xmm0\n" 00049 "\tmovups (%%edi), %%xmm1\n" 00050 "\tmulps %%xmm0, %%xmm1\n" 00051 00052 "\tmovups 16(%%eax), %%xmm5\n" 00053 "\tmovups 16(%%edi), %%xmm6\n" 00054 "\tmulps %%xmm5, %%xmm6\n" 00055 "\taddps %%xmm1, %%xmm3\n" 00056 00057 "\tmovups 32(%%eax), %%xmm0\n" 00058 "\tmovups 32(%%edi), %%xmm1\n" 00059 "\tmulps %%xmm0, %%xmm1\n" 00060 "\taddps %%xmm6, %%xmm4\n" 00061 00062 "\tmovups 48(%%eax), %%xmm5\n" 00063 "\tmovups 48(%%edi), %%xmm6\n" 00064 "\tmulps %%xmm5, %%xmm6\n" 00065 "\taddps %%xmm1, %%xmm3\n" 00066 00067 "\tmovups 64(%%eax), %%xmm0\n" 00068 "\tmovups 64(%%edi), %%xmm1\n" 00069 "\tmulps %%xmm0, %%xmm1\n" 00070 "\taddps %%xmm6, %%xmm4\n" 00071 "\taddps %%xmm1, %%xmm3\n" 00072 00073 00074 "\tadd $80, %%eax\n" 00075 "\tadd $80, %%edi\n" 00076 00077 "\tsub $20, %%ecx\n" 00078 00079 "\tjae .mul20_loop%=\n" 00080 00081 "\taddps %%xmm4, %%xmm3\n" 00082 00083 "\tmovhlps %%xmm3, %%xmm4\n" 00084 "\taddps %%xmm4, %%xmm3\n" 00085 "\tmovaps %%xmm3, %%xmm4\n" 00086 "\tshufps $0x55, %%xmm4, %%xmm4\n" 00087 "\taddss %%xmm4, %%xmm3\n" 00088 "\tmovss %%xmm3, (%%edx)\n" 00089 00090 "\tpop %%ecx\n" 00091 "\tpop %%edi\n" 00092 "\tpop %%eax\n" 00093 : : "a" (a), "D" (b), "c" (len), "d" (&sum) : "memory"); 00094 return sum; 00095 }

Generated on Mon Oct 11 15:42:53 2004 for speex by doxygen 1.3.7