| 1 | /* |
| 2 | * Alpha optimized DSP utils |
| 3 | * Copyright (c) 2002 Falk Hueffner <falk@debian.org> |
| 4 | * |
| 5 | * This file is part of FFmpeg. |
| 6 | * |
| 7 | * FFmpeg is free software; you can redistribute it and/or |
| 8 | * modify it under the terms of the GNU Lesser General Public |
| 9 | * License as published by the Free Software Foundation; either |
| 10 | * version 2.1 of the License, or (at your option) any later version. |
| 11 | * |
| 12 | * FFmpeg is distributed in the hope that it will be useful, |
| 13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 15 | * Lesser General Public License for more details. |
| 16 | * |
| 17 | * You should have received a copy of the GNU Lesser General Public |
| 18 | * License along with FFmpeg; if not, write to the Free Software |
| 19 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 20 | */ |
| 21 | |
| 22 | #include "libavutil/attributes.h" |
| 23 | #include "libavcodec/mpegvideo.h" |
| 24 | #include "asm.h" |
| 25 | |
| 26 | static void dct_unquantize_h263_axp(int16_t *block, int n_coeffs, |
| 27 | uint64_t qscale, uint64_t qadd) |
| 28 | { |
| 29 | uint64_t qmul = qscale << 1; |
| 30 | uint64_t correction = WORD_VEC(qmul * 255 >> 8); |
| 31 | int i; |
| 32 | |
| 33 | qadd = WORD_VEC(qadd); |
| 34 | |
| 35 | for(i = 0; i <= n_coeffs; block += 4, i += 4) { |
| 36 | uint64_t levels, negmask, zeros, add, sub; |
| 37 | |
| 38 | levels = ldq(block); |
| 39 | if (levels == 0) |
| 40 | continue; |
| 41 | |
| 42 | #ifdef __alpha_max__ |
| 43 | /* I don't think the speed difference justifies runtime |
| 44 | detection. */ |
| 45 | negmask = maxsw4(levels, -1); /* negative -> ffff (-1) */ |
| 46 | negmask = minsw4(negmask, 0); /* positive -> 0000 (0) */ |
| 47 | #else |
| 48 | negmask = cmpbge(WORD_VEC(0x7fff), levels); |
| 49 | negmask &= (negmask >> 1) | (1 << 7); |
| 50 | negmask = zap(-1, negmask); |
| 51 | #endif |
| 52 | |
| 53 | zeros = cmpbge(0, levels); |
| 54 | zeros &= zeros >> 1; |
| 55 | /* zeros |= zeros << 1 is not needed since qadd <= 255, so |
| 56 | zapping the lower byte suffices. */ |
| 57 | |
| 58 | levels *= qmul; |
| 59 | levels -= correction & (negmask << 16); |
| 60 | |
| 61 | add = qadd & ~negmask; |
| 62 | sub = qadd & negmask; |
| 63 | /* Set qadd to 0 for levels == 0. */ |
| 64 | add = zap(add, zeros); |
| 65 | levels += add; |
| 66 | levels -= sub; |
| 67 | |
| 68 | stq(levels, block); |
| 69 | } |
| 70 | } |
| 71 | |
| 72 | static void dct_unquantize_h263_intra_axp(MpegEncContext *s, int16_t *block, |
| 73 | int n, int qscale) |
| 74 | { |
| 75 | int n_coeffs; |
| 76 | uint64_t qadd; |
| 77 | int16_t block0 = block[0]; |
| 78 | |
| 79 | if (!s->h263_aic) { |
| 80 | if (n < 4) |
| 81 | block0 *= s->y_dc_scale; |
| 82 | else |
| 83 | block0 *= s->c_dc_scale; |
| 84 | qadd = (qscale - 1) | 1; |
| 85 | } else { |
| 86 | qadd = 0; |
| 87 | } |
| 88 | |
| 89 | if(s->ac_pred) |
| 90 | n_coeffs = 63; |
| 91 | else |
| 92 | n_coeffs = s->inter_scantable.raster_end[s->block_last_index[n]]; |
| 93 | |
| 94 | dct_unquantize_h263_axp(block, n_coeffs, qscale, qadd); |
| 95 | |
| 96 | block[0] = block0; |
| 97 | } |
| 98 | |
| 99 | static void dct_unquantize_h263_inter_axp(MpegEncContext *s, int16_t *block, |
| 100 | int n, int qscale) |
| 101 | { |
| 102 | int n_coeffs = s->inter_scantable.raster_end[s->block_last_index[n]]; |
| 103 | dct_unquantize_h263_axp(block, n_coeffs, qscale, (qscale - 1) | 1); |
| 104 | } |
| 105 | |
| 106 | av_cold void ff_mpv_common_init_axp(MpegEncContext *s) |
| 107 | { |
| 108 | s->dct_unquantize_h263_intra = dct_unquantize_h263_intra_axp; |
| 109 | s->dct_unquantize_h263_inter = dct_unquantize_h263_inter_axp; |
| 110 | } |