Libav 0.7.1
|
00001 /* 00002 * FFT/IFFT transforms 00003 * AltiVec-enabled 00004 * Copyright (c) 2009 Loren Merritt 00005 * 00006 * This file is part of Libav. 00007 * 00008 * Libav is free software; you can redistribute it and/or 00009 * modify it under the terms of the GNU Lesser General Public 00010 * License as published by the Free Software Foundation; either 00011 * version 2.1 of the License, or (at your option) any later version. 00012 * 00013 * Libav is distributed in the hope that it will be useful, 00014 * but WITHOUT ANY WARRANTY; without even the implied warranty of 00015 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 00016 * Lesser General Public License for more details. 00017 * 00018 * You should have received a copy of the GNU Lesser General Public 00019 * License along with Libav; if not, write to the Free Software 00020 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA 00021 */ 00022 #include "libavcodec/fft.h" 00023 #include "util_altivec.h" 00024 #include "types_altivec.h" 00025 00035 void ff_fft_calc_altivec(FFTContext *s, FFTComplex *z); 00036 void ff_fft_calc_interleave_altivec(FFTContext *s, FFTComplex *z); 00037 00038 #if HAVE_GNU_AS 00039 static void ff_imdct_half_altivec(FFTContext *s, FFTSample *output, const FFTSample *input) 00040 { 00041 int j, k; 00042 int n = 1 << s->mdct_bits; 00043 int n4 = n >> 2; 00044 int n8 = n >> 3; 00045 int n32 = n >> 5; 00046 const uint16_t *revtabj = s->revtab; 00047 const uint16_t *revtabk = s->revtab+n4; 00048 const vec_f *tcos = (const vec_f*)(s->tcos+n8); 00049 const vec_f *tsin = (const vec_f*)(s->tsin+n8); 00050 const vec_f *pin = (const vec_f*)(input+n4); 00051 vec_f *pout = (vec_f*)(output+n4); 00052 00053 /* pre rotation */ 00054 k = n32-1; 00055 do { 00056 vec_f cos,sin,cos0,sin0,cos1,sin1,re,im,r0,i0,r1,i1,a,b,c,d; 00057 #define CMULA(p,o0,o1,o2,o3)\ 00058 a = pin[ k*2+p]; /* { z[k].re, z[k].im, z[k+1].re, z[k+1].im } */\ 00059 b = pin[-k*2-p-1]; /* { z[-k-2].re, z[-k-2].im, z[-k-1].re, z[-k-1].im } */\ 00060 re = vec_perm(a, b, vcprm(0,2,s0,s2)); /* { z[k].re, z[k+1].re, z[-k-2].re, z[-k-1].re } */\ 00061 im = vec_perm(a, b, vcprm(s3,s1,3,1)); /* { z[-k-1].im, z[-k-2].im, z[k+1].im, z[k].im } */\ 00062 cos = vec_perm(cos0, cos1, vcprm(o0,o1,s##o2,s##o3)); /* { cos[k], cos[k+1], cos[-k-2], cos[-k-1] } */\ 00063 sin = vec_perm(sin0, sin1, vcprm(o0,o1,s##o2,s##o3));\ 00064 r##p = im*cos - re*sin;\ 00065 i##p = re*cos + im*sin; 00066 #define STORE2(v,dst)\ 00067 j = dst;\ 00068 vec_ste(v, 0, output+j*2);\ 00069 vec_ste(v, 4, output+j*2); 00070 #define STORE8(p)\ 00071 a = vec_perm(r##p, i##p, vcprm(0,s0,0,s0));\ 00072 b = vec_perm(r##p, i##p, vcprm(1,s1,1,s1));\ 00073 c = vec_perm(r##p, i##p, vcprm(2,s2,2,s2));\ 00074 d = vec_perm(r##p, i##p, vcprm(3,s3,3,s3));\ 00075 STORE2(a, revtabk[ p*2-4]);\ 00076 STORE2(b, revtabk[ p*2-3]);\ 00077 STORE2(c, revtabj[-p*2+2]);\ 00078 STORE2(d, revtabj[-p*2+3]); 00079 00080 cos0 = tcos[k]; 00081 sin0 = tsin[k]; 00082 cos1 = tcos[-k-1]; 00083 sin1 = tsin[-k-1]; 00084 CMULA(0, 0,1,2,3); 00085 CMULA(1, 2,3,0,1); 00086 STORE8(0); 00087 STORE8(1); 00088 revtabj += 4; 00089 revtabk -= 4; 00090 k--; 00091 } while(k >= 0); 00092 00093 ff_fft_calc_altivec(s, (FFTComplex*)output); 00094 00095 /* post rotation + reordering */ 00096 j = -n32; 00097 k = n32-1; 00098 do { 00099 vec_f cos,sin,re,im,a,b,c,d; 00100 #define CMULB(d0,d1,o)\ 00101 re = pout[o*2];\ 00102 im = pout[o*2+1];\ 00103 cos = tcos[o];\ 00104 sin = tsin[o];\ 00105 d0 = im*sin - re*cos;\ 00106 d1 = re*sin + im*cos; 00107 00108 CMULB(a,b,j); 00109 CMULB(c,d,k); 00110 pout[2*j] = vec_perm(a, d, vcprm(0,s3,1,s2)); 00111 pout[2*j+1] = vec_perm(a, d, vcprm(2,s1,3,s0)); 00112 pout[2*k] = vec_perm(c, b, vcprm(0,s3,1,s2)); 00113 pout[2*k+1] = vec_perm(c, b, vcprm(2,s1,3,s0)); 00114 j++; 00115 k--; 00116 } while(k >= 0); 00117 } 00118 00119 static void ff_imdct_calc_altivec(FFTContext *s, FFTSample *output, const FFTSample *input) 00120 { 00121 int k; 00122 int n = 1 << s->mdct_bits; 00123 int n4 = n >> 2; 00124 int n16 = n >> 4; 00125 vec_u32 sign = {1U<<31,1U<<31,1U<<31,1U<<31}; 00126 vec_u32 *p0 = (vec_u32*)(output+n4); 00127 vec_u32 *p1 = (vec_u32*)(output+n4*3); 00128 00129 ff_imdct_half_altivec(s, output+n4, input); 00130 00131 for (k = 0; k < n16; k++) { 00132 vec_u32 a = p0[k] ^ sign; 00133 vec_u32 b = p1[-k-1]; 00134 p0[-k-1] = vec_perm(a, a, vcprm(3,2,1,0)); 00135 p1[k] = vec_perm(b, b, vcprm(3,2,1,0)); 00136 } 00137 } 00138 #endif /* HAVE_GNU_AS */ 00139 00140 av_cold void ff_fft_init_altivec(FFTContext *s) 00141 { 00142 #if HAVE_GNU_AS 00143 s->fft_calc = ff_fft_calc_interleave_altivec; 00144 s->imdct_calc = ff_imdct_calc_altivec; 00145 s->imdct_half = ff_imdct_half_altivec; 00146 #endif 00147 }