April 2005 - ffmpeg-devel - lists.ffmpeg.org

MPEG-1/2 decoding benchmarks
by diego＠biurrun.de 25 Jul '05

25 Jul '05

Hi! I have repeated my MPEG-1/2 decoding benchmarks with the latest revisions of mpeg12.c. I used a current CVS snapshot of MPlayer for benchmarking with "-vc ffmpeg12 -quiet -vo null -nosound -benchmark" as common options for all runs. Benchmarks were run on the (framebuffer) console, I stopped all processes except syslogd, klogd and getty while running the benchmarks, no console output was generated. I used mplayer <options> <file> | grep BENCHMARK >> results to record the benchmarking results. I benchmarked separately with the "-lavdopts fast" option turned on. Every benchmark was run 10 times and the best result kept. Since the process is supposed to be deterministic, this should be a fair measurement. I tested revision 1.230 - 1.233 of mpeg12.c, 1.234 had no functional changes. Log messages: 1.230 avoid UPDATE_CACHE() in GET_RL_VLC() if not needed 1.231 optimizing non_intra ac coeff decode 1.232 mpeg2_fast_decode_block_intra() 1.233 avoid clearing block[] if its not used at all The movies I used can be found in http://mplayerhq.hu/~diego/benchmarks/ they were: bomb.mpeg - MPEG-1 176x144 heist60.mpeg - MPEG-1 352x240 dvd.mpeg - MPEG-2 720x576 I tested on two different systems, my desktop and notebook desktop: 500MHz AMD K6-III+ 512MB SD-RAM 133 Debian testing (up to date) gcc 2.95.4 (Debian) notebook: 1500MHz G4 with AltiVec 512MB PC2700 (333MHz) DDR SDRAM Debian testing (up to date) gcc 3.3.5 (Debian) If you have further suggestions or things I should do differently, speak up. I _could_ also test on a Pentium MMX 233, but I doubt the numbers will be very interesting, it's probably better if somebody tries a more modern processor. RESULTS: K6-III: 1.231 gave a slight speedup (2.3%) in some cases, 1.232 a slight performance degradation (1.7%) in some cases, no changes in 1.233. G4: 1.231 gave a slight speedup (1.2%) in some cases, 1.232 a slight improvement (1.2%) in one case, almost no changes in 1.233. Diego Here are the raw numbers: ############### # AMD K6-III+ # ############### 1.230: ~~~~~~ mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.767s VO: 0.002s A: 0.000s Sys: 0.117s = 0.886s BENCHMARK%: VC: 86.5613% VO: 0.2794% A: 0.0000% Sys: 13.1593% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 4.329s VO: 0.008s A: 0.000s Sys: 0.401s = 4.739s BENCHMARK%: VC: 91.3633% VO: 0.1651% A: 0.0000% Sys: 8.4716% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 13.096s VO: 0.005s A: 0.000s Sys: 1.250s = 14.351s BENCHMARK%: VC: 91.2560% VO: 0.0348% A: 0.0000% Sys: 8.7092% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.741s VO: 0.002s A: 0.000s Sys: 0.118s = 0.861s BENCHMARK%: VC: 86.0589% VO: 0.2593% A: 0.0000% Sys: 13.6818% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 4.194s VO: 0.007s A: 0.000s Sys: 0.366s = 4.567s BENCHMARK%: VC: 91.8374% VO: 0.1504% A: 0.0000% Sys: 8.0122% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 12.140s VO: 0.005s A: 0.000s Sys: 1.253s = 13.399s BENCHMARK%: VC: 90.6091% VO: 0.0376% A: 0.0000% Sys: 9.3533% = 100.0000% 1.231: ~~~~~~ mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.757s VO: 0.002s A: 0.000s Sys: 0.120s = 0.880s BENCHMARK%: VC: 86.1142% VO: 0.2701% A: 0.0000% Sys: 13.6157% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 4.256s VO: 0.007s A: 0.000s Sys: 0.372s = 4.636s BENCHMARK%: VC: 91.8181% VO: 0.1570% A: 0.0000% Sys: 8.0249% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 12.772s VO: 0.005s A: 0.000s Sys: 1.255s = 14.032s BENCHMARK%: VC: 91.0222% VO: 0.0346% A: 0.0000% Sys: 8.9433% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.740s VO: 0.003s A: 0.000s Sys: 0.121s = 0.864s BENCHMARK%: VC: 85.7312% VO: 0.3009% A: 0.0000% Sys: 13.9679% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 4.173s VO: 0.007s A: 0.000s Sys: 0.384s = 4.565s BENCHMARK%: VC: 91.4272% VO: 0.1592% A: 0.0000% Sys: 8.4136% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 12.074s VO: 0.005s A: 0.000s Sys: 1.248s = 13.327s BENCHMARK%: VC: 90.5978% VO: 0.0374% A: 0.0000% Sys: 9.3648% = 100.0000% 1.232: ~~~~~~ mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.755s VO: 0.003s A: 0.000s Sys: 0.115s = 0.872s BENCHMARK%: VC: 86.5532% VO: 0.2869% A: 0.0000% Sys: 13.1600% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 4.303s VO: 0.007s A: 0.000s Sys: 0.366s = 4.676s BENCHMARK%: VC: 92.0314% VO: 0.1393% A: 0.0000% Sys: 7.8293% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 13.017s VO: 0.005s A: 0.000s Sys: 1.248s = 14.270s BENCHMARK%: VC: 91.2190% VO: 0.0318% A: 0.0000% Sys: 8.7492% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.750s VO: 0.003s A: 0.000s Sys: 0.115s = 0.867s BENCHMARK%: VC: 86.4108% VO: 0.3270% A: 0.0000% Sys: 13.2623% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 4.235s VO: 0.008s A: 0.000s Sys: 0.367s = 4.610s BENCHMARK%: VC: 91.8804% VO: 0.1675% A: 0.0000% Sys: 7.9521% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 12.319s VO: 0.005s A: 0.000s Sys: 1.245s = 13.569s BENCHMARK%: VC: 90.7899% VO: 0.0354% A: 0.0000% Sys: 9.1747% = 100.0000% 1.233: ~~~~~~ mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.763s VO: 0.002s A: 0.000s Sys: 0.119s = 0.884s mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 4.290s VO: 0.007s A: 0.000s Sys: 0.373s = 4.671s BENCHMARK%: VC: 91.8507% VO: 0.1605% A: 0.0000% Sys: 7.9888% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 12.981s VO: 0.005s A: 0.000s Sys: 1.247s = 14.233s BENCHMARK%: VC: 91.2054% VO: 0.0327% A: 0.0000% Sys: 8.7619% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.751s VO: 0.003s A: 0.000s Sys: 0.119s = 0.873s BENCHMARK%: VC: 85.9708% VO: 0.3464% A: 0.0000% Sys: 13.6828% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 4.276s VO: 0.008s A: 0.000s Sys: 0.362s = 4.646s BENCHMARK%: VC: 92.0346% VO: 0.1673% A: 0.0000% Sys: 7.7981% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 12.291s VO: 0.005s A: 0.000s Sys: 1.246s = 13.542s BENCHMARK%: VC: 90.7621% VO: 0.0362% A: 0.0000% Sys: 9.2017% = 100.0000% ############### # Motorola G4 # ############### 1.230: ~~~~~~ mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.322s VO: 0.001s A: 0.000s Sys: 0.065s = 0.388s BENCHMARK%: VC: 82.9979% VO: 0.3616% A: 0.0000% Sys: 16.6404% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 1.755s VO: 0.003s A: 0.000s Sys: 0.158s = 1.916s BENCHMARK%: VC: 91.5799% VO: 0.1693% A: 0.0000% Sys: 8.2507% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 5.851s VO: 0.003s A: 0.000s Sys: 0.766s = 6.620s BENCHMARK%: VC: 88.3842% VO: 0.0422% A: 0.0000% Sys: 11.5737% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.306s VO: 0.001s A: 0.000s Sys: 0.065s = 0.372s BENCHMARK%: VC: 82.2443% VO: 0.4000% A: 0.0000% Sys: 17.3558% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 1.710s VO: 0.003s A: 0.000s Sys: 0.157s = 1.870s BENCHMARK%: VC: 91.4493% VO: 0.1560% A: 0.0000% Sys: 8.3947% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 5.555s VO: 0.003s A: 0.000s Sys: 0.765s = 6.322s BENCHMARK%: VC: 87.8600% VO: 0.0467% A: 0.0000% Sys: 12.0933% = 100.0000% 1.231: ~~~~~~ mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.318s VO: 0.002s A: 0.000s Sys: 0.065s = 0.384s BENCHMARK%: VC: 82.8190% VO: 0.3904% A: 0.0000% Sys: 16.7905% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark heist60.mpeg ENCHMARKs: VC: 1.752s VO: 0.003s A: 0.000s Sys: 0.162s = 1.917s BENCHMARK%: VC: 91.3908% VO: 0.1795% A: 0.0000% Sys: 8.4297% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 5.813s VO: 0.003s A: 0.000s Sys: 0.764s = 6.580s BENCHMARK%: VC: 88.3431% VO: 0.0473% A: 0.0000% Sys: 11.6096% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.301s VO: 0.001s A: 0.000s Sys: 0.064s = 0.366s BENCHMARK%: VC: 82.0236% VO: 0.3869% A: 0.0000% Sys: 17.5894% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 1.708s VO: 0.003s A: 0.000s Sys: 0.160s = 1.871s BENCHMARK%: VC: 91.2674% VO: 0.1611% A: 0.0000% Sys: 8.5715% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 5.472s VO: 0.003s A: 0.000s Sys: 0.765s = 6.240s BENCHMARK%: VC: 87.6865% VO: 0.0475% A: 0.0000% Sys: 12.2660% = 100.0000% 1.232: ~~~~~~ mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.318s VO: 0.001s A: 0.000s Sys: 0.064s = 0.384s BENCHMARK%: VC: 82.8557% VO: 0.3629% A: 0.0000% Sys: 16.7814% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 1.736s VO: 0.003s A: 0.000s Sys: 0.162s = 1.901s BENCHMARK%: VC: 91.3084% VO: 0.1663% A: 0.0000% Sys: 8.5252% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 5.794s VO: 0.003s A: 0.000s Sys: 0.765s = 6.562s BENCHMARK%: VC: 88.2966% VO: 0.0430% A: 0.0000% Sys: 11.6603% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.300s VO: 0.001s A: 0.000s Sys: 0.065s = 0.366s BENCHMARK%: VC: 81.9985% VO: 0.3761% A: 0.0000% Sys: 17.6254% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 1.689s VO: 0.003s A: 0.000s Sys: 0.157s = 1.849s BENCHMARK%: VC: 91.3421% VO: 0.1492% A: 0.0000% Sys: 8.5087% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 5.455s VO: 0.003s A: 0.000s Sys: 0.765s = 6.223s BENCHMARK%: VC: 87.6639% VO: 0.0464% A: 0.0000% Sys: 12.2897% = 100.0000% 1.233: ~~~~~~ mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.317s VO: 0.001s A: 0.000s Sys: 0.064s = 0.383s BENCHMARK%: VC: 82.7905% VO: 0.3877% A: 0.0000% Sys: 16.8218% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 1.724s VO: 0.003s A: 0.000s Sys: 0.164s = 1.891s BENCHMARK%: VC: 91.1518% VO: 0.1577% A: 0.0000% Sys: 8.6905% = 100.0000% mplayer -vc ffmpeg12 -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 5.769s VO: 0.003s A: 0.000s Sys: 0.770s = 6.542s BENCHMARK%: VC: 88.1854% VO: 0.0475% A: 0.0000% Sys: 11.7671% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark bomb.mpeg BENCHMARKs: VC: 0.299s VO: 0.001s A: 0.000s Sys: 0.064s = 0.365s BENCHMARK%: VC: 81.9638% VO: 0.3892% A: 0.0000% Sys: 17.6469% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark heist60.mpeg BENCHMARKs: VC: 1.686s VO: 0.003s A: 0.000s Sys: 0.160s = 1.849s BENCHMARK%: VC: 91.1841% VO: 0.1595% A: 0.0000% Sys: 8.6565% = 100.0000% mplayer -vc ffmpeg12 -lavdopts fast -quiet -vo null -nosound -benchmark dvd.mpeg BENCHMARKs: VC: 5.421s VO: 0.003s A: 0.000s Sys: 0.767s = 6.191s BENCHMARK%: VC: 87.5587% VO: 0.0478% A: 0.0000% Sys: 12.3935% = 100.0000%

2 3

Bug in ASF mux? (was: RE: [Ffmpeg-user] reduce the time delay)
by dario＠sinistro.net 24 May '05

24 May '05

> > Widows Media Player is delayed 3 seconds. How can I further reduce the > time > > delay. The movie is distribued within LAN, so I do not care the bitrate. > > Please help me. Thank you! > > The Windows Media client adds a delay to allow for buffering. The delay is > configurable within the client settings, but there will always be some > delay due to encoding, propagation and stream setup delay. Different > clients watching the same "Live" stream will often exhibit different > delays. > > Ian Grush > I have noticed that the ASF multiplexer in ffmpeg is not really optmized according to specs. There is something related to preroll header field, and timestamps. A normal ASF file (there a really nice utility called Windows Media ASF Viewer, http://www.microsoft.com/windows/windowsmedia/format/ASFViewer.aspx, that may be used to examine the ASF file further) has a preroll equal to the buffer window (normally 3 or 5 secs), and timestamps (actually packet send times) beginning exactly at the specified preroll (from the header field). >From the files generated by ffmpeg, I could see there are some glitches (such as local playback delay in the windows media player, not the one created by the network buffering, etc...). Is anyone willing to fix it? Regards, hope it helps, Dario Andrade

2 1

Spelling error:-P
by kevinmb500＠gawab.com 24 May '05

24 May '05

i found a spelling error in mpegvideo.c It prints out sometimes when encoding msmpeg4, msmpegv2, wmv1, wmv2. I'm a pretty lousy coder...maybe i'll spend my time doing insignificant stuff like this. :D

12 30

ac3enc.c modifications
by jruggle＠earthlink.net 24 May '05

24 May '05

Hello, Here are some of the additions I've made to the ac3 encoder. I've added stereo rematrixing and channel gain range. Stereo rematrixing selectively uses [(L+R)/2, (L-R)/2] instead of [L, R] when the 2 channels are similar. Channel gain range is used in the decoder to increase dynamic range. The only thing I've changed (rather than added) is on line 1404. I changed v-8 to v-9 because I noticed that the output was too quiet. Using v-8, the exponent is limited to 1-24 instead of 0-24. Changing to v-9 fixes this. Let me know what you think. -Justin --- ac3enc-orig.c 2005-04-20 20:04:26.000000000 -0400 +++ ac3enc.c 2005-04-20 20:10:05.000000000 -0400 @@ -967,6 +967,9 @@ uint8_t bap[AC3_MAX_CHANNELS][N/2], int32_t mdct_coefs[AC3_MAX_CHANNELS][N/2], int8_t global_exp[AC3_MAX_CHANNELS], + uint8_t gainrng[AC3_MAX_CHANNELS], + uint8_t rematstr, + uint8_t rematflg[4], int block_num) { int ch, nb_groups, group_size, i, baie, rbnd; @@ -991,23 +994,16 @@ put_bits(&s->pb, 1, 0); /* no new coupling strategy */ } - if (s->acmod == 2) - { - if(block_num==0) - { - /* first block must define rematrixing (rematstr) */ - put_bits(&s->pb, 1, 1); - - /* dummy rematrixing rematflg(1:4)=0 */ - for (rbnd=0;rbnd<4;rbnd++) - put_bits(&s->pb, 1, 0); - } - else - { - /* no matrixing (but should be used in the future) */ - put_bits(&s->pb, 1, 0); - } - } + if(s->acmod == 2) { + // rematrixing strategy + put_bits(&s->pb, 1, rematstr); + + // rematrixing flags + if(rematstr) { + for(rbnd=0; rbnd<4; rbnd++) + put_bits(&s->pb, 1, rematflg[rbnd]); + } + } #if defined(DEBUG) { @@ -1074,7 +1070,7 @@ } if (ch != s->lfe_channel) - put_bits(&s->pb, 2, 0); /* no gain range info */ + put_bits(&s->pb, 2, gainrng[ch]); /* gain range info */ } /* bit allocation info */ @@ -1355,6 +1351,69 @@ return frame_size * 2; } + +/* + * channel gain range + */ +static uint8_t +calc_ch_gain_range(int16_t *tab, int n) { + int i; + int v = 0; + for(i=0; i<n; i++) { + v |= abs(tab[i]); + } + uint8_t gain = 0; + if(v >= 16384) gain = 0; + else if(v >= 8192) gain = 1; + else if(v >= 4096) gain = 2; + else gain = 3; + return gain; +} + + +/* probably should go in ac3tab.h */ +static uint8_t rematbndtab[4][2] = { {13, 24}, {25, 36}, {37, 60}, {61, 252} }; + +static void calc_rematrixing(int32_t mdct_coef[AC3_MAX_CHANNELS][N/2], + uint8_t *rematstr, uint8_t rematflg[NB_BLOCKS][4], int blk) +{ + uint32_t sum[4][4]; + int32_t lt, rt, ctmp; + int bnd, i, chmin; + + memset(sum, 0, 64); + *rematstr = 0; + for(bnd=0; bnd<4; bnd++) { + for(i=rematbndtab[bnd][0]; i<=rematbndtab[bnd][1]; i++) { + lt = mdct_coef[0][i] / 128; + rt = mdct_coef[1][i] / 128; + sum[bnd][0] += lt*lt; + sum[bnd][1] += rt*rt; + sum[bnd][2] += (lt+rt)*(lt+rt); + sum[bnd][3] += (lt-rt)*(lt-rt); + } + chmin = 0; + if(sum[bnd][1] < sum[bnd][chmin]) chmin = 1; + if(sum[bnd][2] < sum[bnd][chmin]) chmin = 2; + if(sum[bnd][3] < sum[bnd][chmin]) chmin = 3; + if(chmin<2) { + rematflg[blk][bnd] = 0; + } + else { + rematflg[blk][bnd] = 1; + for(i=rematbndtab[bnd][0]; i<=rematbndtab[bnd][1]; i++) { + ctmp = mdct_coef[0][i]; + mdct_coef[0][i] = (ctmp + mdct_coef[1][i]) / 2; + mdct_coef[1][i] = (ctmp - mdct_coef[1][i]) / 2; + } + } + if(blk != 0 && *rematstr == 0 && + rematflg[blk][bnd] != rematflg[blk-1][bnd]) { + *rematstr = 1; + } + } + if(blk == 0) *rematstr = 1; +} + static int AC3_encode_frame(AVCodecContext *avctx, unsigned char *frame, int buf_size, void *data) { @@ -1368,6 +1427,9 @@ uint8_t encoded_exp[NB_BLOCKS][AC3_MAX_CHANNELS][N/2]; uint8_t bap[NB_BLOCKS][AC3_MAX_CHANNELS][N/2]; int8_t exp_samples[NB_BLOCKS][AC3_MAX_CHANNELS]; + uint8_t gainrng[NB_BLOCKS][AC3_MAX_CHANNELS]; + uint8_t rematstr[NB_BLOCKS]; + uint8_t rematflg[NB_BLOCKS][4]; int frame_bits; frame_bits = 0; @@ -1388,6 +1450,9 @@ sptr += sinc; } + /* calculate channel gain range */ + gainrng[i][ch] = calc_ch_gain_range(input_samples, 512); + /* apply the MDCT window */ for(j=0;j<N/2;j++) { input_samples[j] = MUL16(input_samples[j], @@ -1401,12 +1466,24 @@ v = 14 - log2_tab(input_samples, N); if (v < 0) v = 0; - exp_samples[i][ch] = v - 8; + exp_samples[i][ch] = v - 9; lshift_tab(input_samples, N, v); /* do the MDCT */ mdct512(mdct_coef[i][ch], input_samples); - + } + } + + // calculate rematrixing + if(s->acmod == 2) { + for(i=0;i<NB_BLOCKS;i++) { + calc_rematrixing(mdct_coef[i], &rematstr[i], rematflg, i); + } + } + + + for(ch=0;ch<s->nb_all_channels;ch++) { + for(i=0;i<NB_BLOCKS;i++) { /* compute "exponents". We take into account the normalization there */ for(j=0;j<N/2;j++) { @@ -1455,7 +1532,8 @@ for(i=0;i<NB_BLOCKS;i++) { output_audio_block(s, exp_strategy[i], encoded_exp[i], - bap[i], mdct_coef[i], exp_samples[i], i); + bap[i], mdct_coef[i], exp_samples[i], + gainrng[i], rematstr[i], rematflg[i], i); } return output_frame_end(s); }

12 44

segfault in ff_mpeg1_find_frame_end()
by stefan＠lucke.in-berlin.de 14 May '05

14 May '05

Hi, We are developing a plugin for vdr named softdevice. Plugins homepage is at http://softdevice.berlios.de . I get a segfault in function mentioned in subject. The segfault happens when there is a transition in aspect ratio from 16:9 to 4:3. It happens not when the stream is played the first time but when its played the 3rd time. But this may vary (segfault may happen at the first time too). To ensure that it has nothing to with PADDING bytes I added extra code that mallocs and copies data before decoding and zeros from "data+size" FF_INPUT_BUFFER_PADDING_SIZE bytes. I'm using ffmpeg cvs version from today. What could be done to track down the cause of segfault? Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 655401 (LWP 4847)] 0x4046d1d1 in ff_mpeg1_find_frame_end () from /usr/local/lib/libavcodec.so (gdb) bt #0 0x4046d1d1 in ff_mpeg1_find_frame_end () from /usr/local/lib/libavcodec.so #1 0x4046c8a9 in ff_mpeg1_find_frame_end () from /usr/local/lib/libavcodec.so #2 0x40373154 in avcodec_decode_video () from /usr/local/lib/libavcodec.so #3 0x403099f9 in cVideoStreamDecoder::DecodePacket(AVPacket*) (this=0x841f820, pkt=0x8423a24) at mpeg2decoder.c:497 #4 0x40308b29 in cStreamDecoder::Action() (this=0x841f820) at mpeg2decoder.c:199 #5 0x080dbc40 in cThread::StartThread(cThread*) (Thread=0x841f820) at thread.c:227 #6 0x40059f60 in pthread_start_thread () from /lib/i686/libpthread.so.0 #7 0x4005a0fe in pthread_start_thread_event () from /lib/i686/libpthread.so.0 #8 0x40270327 in clone () from /lib/i686/libc.so.6 and the following at the first try when using statically linked ffmpegs libs: Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 245776 (LWP 5433)] mpeg1_decode_picture (avctx=0x8368fc8, buf=0x86dc258 <Address 0x86dc258 out of bounds>, buf_size=18921) at bswap.h:30 30 { (gdb) bt #0 mpeg1_decode_picture (avctx=0x8368fc8, buf=0x86dc258 <Address 0x86dc258 out of bounds>, buf_size=18921) at bswap.h:30 #1 0x4045a799 in mpeg_decode_frame (avctx=0x8368fc8, data=0x8451340, data_size=0x489dfe54, buf=0x86dc190 <Address 0x86dc190 out of bounds>, buf_size=19121) at mpeg12.c:3023 #2 0x40364fc4 in avcodec_decode_video (avctx=0x8368fc8, picture=0x8451340, got_picture_ptr=0x489dfe54, buf=0x8432180 "", buf_size=2039) at utils.c:593 #3 0x4035f199 in cVideoStreamDecoder::DecodePacket(AVPacket*) (this=0x8440b48, pkt=0x8444d4c) at mpeg2decoder.c:497 #4 0x4035e2c9 in cStreamDecoder::Action() (this=0x8440b48) at mpeg2decoder.c:199 #5 0x080dbc40 in cThread::StartThread(cThread*) (Thread=0x8440b48) at thread.c:227 #6 0x40059f60 in pthread_start_thread () from /lib/i686/libpthread.so.0 #7 0x4005a0fe in pthread_start_thread_event () from /lib/i686/libpthread.so.0 #8 0x40270327 in clone () from /lib/i686/libc.so.6 -- Stefan Lucke

4 20

[Ffmpeg-devel-old] Native vorbis decoder, unfinished. (was: Re: [Ffmpeg-devel] Native ogg demuxer)
by dbalatoni＠programozo.hu 13 May '05

13 May '05

Hi! 2005. ?prilis 11. 18.12-kor Alex Beregszaszi ezeket a bolcs gondolatokat fogalmazta meg: > Attached is my my start on a vorbis implementation, but sadly it looks > like I wont have time to finish it (pending since last year). > > It has most of the setup packet decoding in it, the actual frame > decoding is a much easier step. No. 1.) you have to build the vlc table 2.) you have to decode the floor 3.) you have to decode the residue 4.) inverse coupling 5.) you have to take the dot product of the floor and the residue 6.) (mdct +)windowing It might sound easy, but 1.), 2.) and 6.) proved tricky for me (and 3. was/is wrong in the full html version of the specification). > Alex Beregszaszi e-mail: alex at fsn.hu > Free Software Network cell: +36 70 3144424 bye Denes -- - Use the Source Luke ! -

4 15

ac3enc.c bit allocation fix
by jruggle＠earthlink.net 05 May '05

05 May '05

Hello, While adding stereo rematrixing, I came across something that needs to be fixed even without adding the feature. The output correctly uses 4 dummy values for the rematrixing flags in block-0, but the bit allocation routine does not take these bits into account. From what I can tell, there was a patch in 2003 that corrected the output to make it DVD and spec compatible, but it didn't correct the bit allocation. It's only 4 bits over the entire 6 blocks, so overflow errors would happen rarely or never, but it's still worth fixing. So here is a fix. -Justin

2 1

[Ffmpeg-devel-old] Crash with SSE compare function
by milan_cutka＠yahoo.com 05 May '05

05 May '05

Hi, ffmpeg crashes when encoding to MPEG4 with following command line: ffmpeg.exe -i 1.avi -cmp 257 -4mv 2.avi The crash occurs in dsputils.c in function sse4_c with pix2 parameter being NULL. Sorry for not sending GDB backtrace, but I wasn't able to get one in mingw environment. Milan ------------------------------------------------------- SF.Net email is sponsored by: Tell us your software development plans! Take this survey and enter to win a one-year sub to SourceForge.net Plus IDC's 2005 look-ahead and a copy of this survey Click here to start! http://www.idcswdc.com/cgi-bin/survey?id=105hix _______________________________________________ Ffmpeg-devel mailing list Ffmpeg-devel at lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ffmpeg-devel

2 2

Revised os2.diff
by paul＠smedley.info 05 May '05

05 May '05

>>+ifeq ($(CONFIG_OS2),yes) >>+ $(CC) $(SHFLAGS) -o $@ $(OBJS) $(PPOBJS) $(AVCLIBS) $(EXTRALIBS) $(LDFLAGS) >> else >> $(CC) $(SHFLAGS) -o $@ $(OBJS) $(PPOBJS) $(AVCLIBS) $(EXTRALIBS) >> endif > > > Please edit the configure script and add the required flags to the SHFLAGS. > > lu Done. Revised os2.diff attached. Cheers, Paul.

6 5

ffmpeg compile fail w/gcc 3.4.3 due to lrintf definition
by rlippmann＠imsventures.com 03 May '05

03 May '05

hi all, not sure if this has been covered (can't seem to find the old devel mailing list archives to search, they're gone from sourceforge), but: tried to compile today's CVS using fedora core 3 with latest updates, and I fail with the following: make -C libavcodec all make[1]: Entering directory `/home/rlippmann/ffmpeg/libavcodec' gcc -O3 -g -Wall -Wno-switch -DHAVE_AV_CONFIG_H -I.. -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -D_GNU_SOURCE -c -o bitstream.o bitstream.c gcc -O3 -g -Wall -Wno-switch -DHAVE_AV_CONFIG_H -I.. -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -D_GNU_SOURCE -c -o utils.o utils.c In file included from utils.c:28: dsputil.h:617: warning: static declaration of 'lrintf' follows non-static declaration gcc -O3 -g -Wall -Wno-switch -DHAVE_AV_CONFIG_H -I.. -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -D_GNU_SOURCE -c -o mem.o mem.c gcc -O3 -g -Wall -Wno-switch -DHAVE_AV_CONFIG_H -I.. -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -D_GNU_SOURCE -c -o allcodecs.o allcodecs.c gcc -O3 -g -Wall -Wno-switch -DHAVE_AV_CONFIG_H -I.. -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -D_GNU_SOURCE -c -o mpegvideo.o mpegvideo.c In file included from mpegvideo.c:29: dsputil.h:617: warning: static declaration of 'lrintf' follows non-static declaration mpegvideo.c: In function `MPV_encode_picture': dsputil.h:617: sorry, unimplemented: inlining failed in call to 'lrintf': redefined extern inline functions are not considered for inlining mpegvideo.c:2394: sorry, unimplemented: called from here make[1]: *** [mpegvideo.o] Error 1 make[1]: Leaving directory `/home/rlippmann/ffmpeg/libavcodec' make: *** [lib] Error 2 configure was invoked as follows: ./configure --disable-ffplay --disable-v4l --disable-dv1394 --disable-audio -beos --enable-pthreads --enable-gpl --enable-dts --enable-a52 --enable-faad --enable-faadbin gcc version: Reading specs from /usr/lib/gcc/i386-redhat-linux/3.4.3/specs Configured with: ../configure --prefix=/usr --mandir=/usr/share/man --infodir=/usr/share/info --enable-shared --enable-threads=posix --disable-checking --with-system-zli b --enable-__cxa_atexit --disable-libunwind-exceptions --enable-java-awt=gtk --host=i386-redhat-linux Thread model: posix gcc version 3.4.3 20050227 (Red Hat 3.4.3-22.fc3) any ideas? thanks, rob

2 2