/* * Copyright (c) 2012 * MIPS Technologies, Inc., California. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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. * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its * contributors may be used to endorse or promote products derived from * this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``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 MIPS TECHNOLOGIES, INC. 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. * * Author: Stanislav Ocovaj (socovaj@mips.com) * Author: Zoran Lukic (zoranl@mips.com) * * Optimized MDCT/IMDCT and FFT transforms * * This file is part of FFmpeg. * * FFmpeg is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * * FFmpeg is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with FFmpeg; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ #include "config.h" #include "libavcodec/fft.h" #include "libavcodec/fft_table.h" #include "libavutil/mips/asmdefs.h" /** * FFT transform */ #if HAVE_INLINE_ASM #if !HAVE_MIPS32R6 && !HAVE_MIPS64R6 static void ff_fft_calc_mips(FFTContext *s, FFTComplex *z) { int nbits, i, n, num_transforms, offset, step; int n4, n2, n34; FFTSample tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8; FFTComplex *tmpz; float w_re, w_im; float *w_re_ptr, *w_im_ptr; const int fft_size = (1 << s->nbits); float pom, pom1, pom2, pom3; float temp, temp1, temp3, temp4; FFTComplex * tmpz_n2, * tmpz_n34, * tmpz_n4; FFTComplex * tmpz_n2_i, * tmpz_n34_i, * tmpz_n4_i, * tmpz_i; num_transforms = (21845 >> (17 - s->nbits)) | 1; for (n=0; n> 1) | 1; for (n=0; nnbits; nbits++) { num_transforms = (num_transforms >> 1) | 1; n2 = 2 * n4; n34 = 3 * n4; for (n=0; n>= 1; n4 <<= 1; } } /** * MDCT/IMDCT transforms. */ static void ff_imdct_half_mips(FFTContext *s, FFTSample *output, const FFTSample *input) { int k, n8, n4, n2, n, j; const uint16_t *revtab = s->revtab; const FFTSample *tcos = s->tcos; const FFTSample *tsin = s->tsin; const FFTSample *in1, *in2, *in3, *in4; FFTComplex *z = (FFTComplex *)output; int j1; const float *tcos1, *tsin1, *tcos2, *tsin2; float temp1, temp2, temp3, temp4, temp5, temp6, temp7, temp8, temp9, temp10, temp11, temp12, temp13, temp14, temp15, temp16; FFTComplex *z1, *z2; n = 1 << s->mdct_bits; n2 = n >> 1; n4 = n >> 2; n8 = n >> 3; /* pre rotation */ in1 = input; in2 = input + n2 - 1; in3 = input + 2; in4 = input + n2 - 3; tcos1 = tcos; tsin1 = tsin; /* n4 = 64 or 128 */ for(k = 0; k < n4; k += 2) { j = revtab[k ]; j1 = revtab[k + 1]; __asm__ volatile ( "lwc1 %[temp1], 0(%[in2]) \t\n" "lwc1 %[temp2], 0(%[tcos1]) \t\n" "lwc1 %[temp3], 0(%[tsin1]) \t\n" "lwc1 %[temp4], 0(%[in1]) \t\n" "lwc1 %[temp5], 0(%[in4]) \t\n" "mul.s %[temp9], %[temp1], %[temp2] \t\n" "mul.s %[temp10], %[temp1], %[temp3] \t\n" "lwc1 %[temp6], 4(%[tcos1]) \t\n" "lwc1 %[temp7], 4(%[tsin1]) \t\n" "nmsub.s %[temp9], %[temp9], %[temp4], %[temp3] \t\n" "madd.s %[temp10], %[temp10], %[temp4], %[temp2] \t\n" "mul.s %[temp11], %[temp5], %[temp6] \t\n" "mul.s %[temp12], %[temp5], %[temp7] \t\n" "lwc1 %[temp8], 0(%[in3]) \t\n" PTR_ADDIU " %[tcos1], %[tcos1], 8 \t\n" PTR_ADDIU " %[tsin1], %[tsin1], 8 \t\n" PTR_ADDIU " %[in1], %[in1], 16 \t\n" "nmsub.s %[temp11], %[temp11], %[temp8], %[temp7] \t\n" "madd.s %[temp12], %[temp12], %[temp8], %[temp6] \t\n" PTR_ADDIU " %[in2], %[in2], -16 \t\n" PTR_ADDIU " %[in3], %[in3], 16 \t\n" PTR_ADDIU " %[in4], %[in4], -16 \t\n" : [temp1]"=&f"(temp1), [temp2]"=&f"(temp2), [temp3]"=&f"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5), [temp6]"=&f"(temp6), [temp7]"=&f"(temp7), [temp8]"=&f"(temp8), [temp9]"=&f"(temp9), [temp10]"=&f"(temp10), [temp11]"=&f"(temp11), [temp12]"=&f"(temp12), [tsin1]"+r"(tsin1), [tcos1]"+r"(tcos1), [in1]"+r"(in1), [in2]"+r"(in2), [in3]"+r"(in3), [in4]"+r"(in4) : : "memory" ); z[j ].re = temp9; z[j ].im = temp10; z[j1].re = temp11; z[j1].im = temp12; } s->fft_calc(s, z); /* post rotation + reordering */ /* n8 = 32 or 64 */ for(k = 0; k < n8; k += 2) { tcos1 = &tcos[n8 - k - 2]; tsin1 = &tsin[n8 - k - 2]; tcos2 = &tcos[n8 + k]; tsin2 = &tsin[n8 + k]; z1 = &z[n8 - k - 2]; z2 = &z[n8 + k ]; __asm__ volatile ( "lwc1 %[temp1], 12(%[z1]) \t\n" "lwc1 %[temp2], 4(%[tsin1]) \t\n" "lwc1 %[temp3], 4(%[tcos1]) \t\n" "lwc1 %[temp4], 8(%[z1]) \t\n" "lwc1 %[temp5], 4(%[z1]) \t\n" "mul.s %[temp9], %[temp1], %[temp2] \t\n" "mul.s %[temp10], %[temp1], %[temp3] \t\n" "lwc1 %[temp6], 0(%[tsin1]) \t\n" "lwc1 %[temp7], 0(%[tcos1]) \t\n" "nmsub.s %[temp9], %[temp9], %[temp4], %[temp3] \t\n" "madd.s %[temp10], %[temp10], %[temp4], %[temp2] \t\n" "mul.s %[temp11], %[temp5], %[temp6] \t\n" "mul.s %[temp12], %[temp5], %[temp7] \t\n" "lwc1 %[temp8], 0(%[z1]) \t\n" "lwc1 %[temp1], 4(%[z2]) \t\n" "lwc1 %[temp2], 0(%[tsin2]) \t\n" "lwc1 %[temp3], 0(%[tcos2]) \t\n" "nmsub.s %[temp11], %[temp11], %[temp8], %[temp7] \t\n" "madd.s %[temp12], %[temp12], %[temp8], %[temp6] \t\n" "mul.s %[temp13], %[temp1], %[temp2] \t\n" "mul.s %[temp14], %[temp1], %[temp3] \t\n" "lwc1 %[temp4], 0(%[z2]) \t\n" "lwc1 %[temp5], 12(%[z2]) \t\n" "lwc1 %[temp6], 4(%[tsin2]) \t\n" "lwc1 %[temp7], 4(%[tcos2]) \t\n" "nmsub.s %[temp13], %[temp13], %[temp4], %[temp3] \t\n" "madd.s %[temp14], %[temp14], %[temp4], %[temp2] \t\n" "mul.s %[temp15], %[temp5], %[temp6] \t\n" "mul.s %[temp16], %[temp5], %[temp7] \t\n" "lwc1 %[temp8], 8(%[z2]) \t\n" "nmsub.s %[temp15], %[temp15], %[temp8], %[temp7] \t\n" "madd.s %[temp16], %[temp16], %[temp8], %[temp6] \t\n" : [temp1]"=&f"(temp1), [temp2]"=&f"(temp2), [temp3]"=&f"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5), [temp6]"=&f"(temp6), [temp7]"=&f"(temp7), [temp8]"=&f"(temp8), [temp9]"=&f"(temp9), [temp10]"=&f"(temp10), [temp11]"=&f"(temp11), [temp12]"=&f"(temp12), [temp13]"=&f"(temp13), [temp14]"=&f"(temp14), [temp15]"=&f"(temp15), [temp16]"=&f"(temp16) : [z1]"r"(z1), [z2]"r"(z2), [tsin1]"r"(tsin1), [tcos1]"r"(tcos1), [tsin2]"r"(tsin2), [tcos2]"r"(tcos2) : "memory" ); z1[1].re = temp9; z1[1].im = temp14; z2[0].re = temp13; z2[0].im = temp10; z1[0].re = temp11; z1[0].im = temp16; z2[1].re = temp15; z2[1].im = temp12; } } /** * Compute inverse MDCT of size N = 2^nbits * @param output N samples * @param input N/2 samples */ static void ff_imdct_calc_mips(FFTContext *s, FFTSample *output, const FFTSample *input) { int k; int n = 1 << s->mdct_bits; int n2 = n >> 1; int n4 = n >> 2; ff_imdct_half_mips(s, output+n4, input); for(k = 0; k < n4; k+=4) { output[k] = -output[n2-k-1]; output[k+1] = -output[n2-k-2]; output[k+2] = -output[n2-k-3]; output[k+3] = -output[n2-k-4]; output[n-k-1] = output[n2+k]; output[n-k-2] = output[n2+k+1]; output[n-k-3] = output[n2+k+2]; output[n-k-4] = output[n2+k+3]; } } #endif /* !HAVE_MIPS32R6 && !HAVE_MIPS64R6 */ #endif /* HAVE_INLINE_ASM */ av_cold void ff_fft_init_mips(FFTContext *s) { int n=0; ff_fft_lut_init(ff_fft_offsets_lut, 0, 1 << 17, &n); ff_init_ff_cos_tabs(17); #if HAVE_INLINE_ASM #if !HAVE_MIPS32R6 && !HAVE_MIPS64R6 s->fft_calc = ff_fft_calc_mips; #if CONFIG_MDCT s->imdct_calc = ff_imdct_calc_mips; s->imdct_half = ff_imdct_half_mips; #endif #endif #endif }