Mercurial > cgi-bin > hgwebdir.cgi > PR > Applications > VSs > VSs__H264__App
comparison libavcodec/arm/mpegvideo_iwmmxt.c @ 9:ea1ba68cf0ed
update to match api changes + add sscc produced source
| author | Nina Engelhardt <nengel@mailbox.tu-berlin.de> |
|---|---|
| date | Wed, 05 Jun 2013 14:43:26 +0200 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| -1:000000000000 | 0:d0ed8756e5aa |
|---|---|
| 1 /* | |
| 2 * copyright (c) 2004 AGAWA Koji | |
| 3 * | |
| 4 * This file is part of FFmpeg. | |
| 5 * | |
| 6 * FFmpeg is free software; you can redistribute it and/or | |
| 7 * modify it under the terms of the GNU Lesser General Public | |
| 8 * License as published by the Free Software Foundation; either | |
| 9 * version 2.1 of the License, or (at your option) any later version. | |
| 10 * | |
| 11 * FFmpeg is distributed in the hope that it will be useful, | |
| 12 * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
| 13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
| 14 * Lesser General Public License for more details. | |
| 15 * | |
| 16 * You should have received a copy of the GNU Lesser General Public | |
| 17 * License along with FFmpeg; if not, write to the Free Software | |
| 18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA | |
| 19 */ | |
| 20 | |
| 21 #include "libavcodec/avcodec.h" | |
| 22 #include "libavcodec/dsputil.h" | |
| 23 #include "libavcodec/mpegvideo.h" | |
| 24 #include "mpegvideo_arm.h" | |
| 25 | |
| 26 static void dct_unquantize_h263_intra_iwmmxt(MpegEncContext *s, | |
| 27 DCTELEM *block, int n, int qscale) | |
| 28 { | |
| 29 int level, qmul, qadd; | |
| 30 int nCoeffs; | |
| 31 DCTELEM *block_orig = block; | |
| 32 | |
| 33 assert(s->block_last_index[n]>=0); | |
| 34 | |
| 35 qmul = qscale << 1; | |
| 36 | |
| 37 if (!s->h263_aic) { | |
| 38 if (n < 4) | |
| 39 level = block[0] * s->y_dc_scale; | |
| 40 else | |
| 41 level = block[0] * s->c_dc_scale; | |
| 42 qadd = (qscale - 1) | 1; | |
| 43 }else{ | |
| 44 qadd = 0; | |
| 45 level = block[0]; | |
| 46 } | |
| 47 if(s->ac_pred) | |
| 48 nCoeffs=63; | |
| 49 else | |
| 50 nCoeffs= s->inter_scantable.raster_end[ s->block_last_index[n] ]; | |
| 51 | |
| 52 __asm__ volatile ( | |
| 53 /* "movd %1, %%mm6 \n\t" //qmul */ | |
| 54 /* "packssdw %%mm6, %%mm6 \n\t" */ | |
| 55 /* "packssdw %%mm6, %%mm6 \n\t" */ | |
| 56 "tbcsth wr6, %[qmul] \n\t" | |
| 57 /* "movd %2, %%mm5 \n\t" //qadd */ | |
| 58 /* "packssdw %%mm5, %%mm5 \n\t" */ | |
| 59 /* "packssdw %%mm5, %%mm5 \n\t" */ | |
| 60 "tbcsth wr5, %[qadd] \n\t" | |
| 61 "wzero wr7 \n\t" /* "pxor %%mm7, %%mm7 \n\t" */ | |
| 62 "wzero wr4 \n\t" /* "pxor %%mm4, %%mm4 \n\t" */ | |
| 63 "wsubh wr7, wr5, wr7 \n\t" /* "psubw %%mm5, %%mm7 \n\t" */ | |
| 64 "1: \n\t" | |
| 65 "wldrd wr2, [%[block]] \n\t" /* "movq (%0, %3), %%mm0 \n\t" */ | |
| 66 "wldrd wr3, [%[block], #8] \n\t" /* "movq 8(%0, %3), %%mm1 \n\t" */ | |
| 67 "wmulsl wr0, wr6, wr2 \n\t" /* "pmullw %%mm6, %%mm0 \n\t" */ | |
| 68 "wmulsl wr1, wr6, wr3 \n\t" /* "pmullw %%mm6, %%mm1 \n\t" */ | |
| 69 /* "movq (%0, %3), %%mm2 \n\t" */ | |
| 70 /* "movq 8(%0, %3), %%mm3 \n\t" */ | |
| 71 "wcmpgtsh wr2, wr4, wr2 \n\t" /* "pcmpgtw %%mm4, %%mm2 \n\t" // block[i] < 0 ? -1 : 0 */ | |
| 72 "wcmpgtsh wr3, wr4, wr2 \n\t" /* "pcmpgtw %%mm4, %%mm3 \n\t" // block[i] < 0 ? -1 : 0 */ | |
| 73 "wxor wr0, wr2, wr0 \n\t" /* "pxor %%mm2, %%mm0 \n\t" */ | |
| 74 "wxor wr1, wr3, wr1 \n\t" /* "pxor %%mm3, %%mm1 \n\t" */ | |
| 75 "waddh wr0, wr7, wr0 \n\t" /* "paddw %%mm7, %%mm0 \n\t" */ | |
| 76 "waddh wr1, wr7, wr1 \n\t" /* "paddw %%mm7, %%mm1 \n\t" */ | |
| 77 "wxor wr2, wr0, wr2 \n\t" /* "pxor %%mm0, %%mm2 \n\t" */ | |
| 78 "wxor wr3, wr1, wr3 \n\t" /* "pxor %%mm1, %%mm3 \n\t" */ | |
| 79 "wcmpeqh wr0, wr7, wr0 \n\t" /* "pcmpeqw %%mm7, %%mm0 \n\t" // block[i] == 0 ? -1 : 0 */ | |
| 80 "wcmpeqh wr1, wr7, wr1 \n\t" /* "pcmpeqw %%mm7, %%mm1 \n\t" // block[i] == 0 ? -1 : 0 */ | |
| 81 "wandn wr0, wr2, wr0 \n\t" /* "pandn %%mm2, %%mm0 \n\t" */ | |
| 82 "wandn wr1, wr3, wr1 \n\t" /* "pandn %%mm3, %%mm1 \n\t" */ | |
| 83 "wstrd wr0, [%[block]] \n\t" /* "movq %%mm0, (%0, %3) \n\t" */ | |
| 84 "wstrd wr1, [%[block], #8] \n\t" /* "movq %%mm1, 8(%0, %3) \n\t" */ | |
| 85 "add %[block], %[block], #16 \n\t" /* "addl $16, %3 \n\t" */ | |
| 86 "subs %[i], %[i], #1 \n\t" | |
| 87 "bne 1b \n\t" /* "jng 1b \n\t" */ | |
| 88 :[block]"+r"(block) | |
| 89 :[i]"r"((nCoeffs + 8) / 8), [qmul]"r"(qmul), [qadd]"r"(qadd) | |
| 90 :"memory"); | |
| 91 | |
| 92 block_orig[0] = level; | |
| 93 } | |
| 94 | |
| 95 #if 0 | |
| 96 static void dct_unquantize_h263_inter_iwmmxt(MpegEncContext *s, | |
| 97 DCTELEM *block, int n, int qscale) | |
| 98 { | |
| 99 int nCoeffs; | |
| 100 | |
| 101 assert(s->block_last_index[n]>=0); | |
| 102 | |
| 103 if(s->ac_pred) | |
| 104 nCoeffs=63; | |
| 105 else | |
| 106 nCoeffs= s->inter_scantable.raster_end[ s->block_last_index[n] ]; | |
| 107 | |
| 108 ippiQuantInvInter_Compact_H263_16s_I(block, nCoeffs+1, qscale); | |
| 109 } | |
| 110 #endif | |
| 111 | |
| 112 void MPV_common_init_iwmmxt(MpegEncContext *s) | |
| 113 { | |
| 114 if (!(mm_flags & FF_MM_IWMMXT)) return; | |
| 115 | |
| 116 s->dct_unquantize_h263_intra = dct_unquantize_h263_intra_iwmmxt; | |
| 117 #if 0 | |
| 118 s->dct_unquantize_h263_inter = dct_unquantize_h263_inter_iwmmxt; | |
| 119 #endif | |
| 120 } |
