Re: [Freevo-users] Help with benchmarking CPUs for HD decoding
Ye :) Thank you Jason. Indeed it was the -vo gl that was causing the frame drops. the -vo xv works like a breeze (<40% of one CPU) I don't know why I choose it and did not think to test over vo... but this was my mistake. Thank you again... So for future googlers: 1080p IS working nice and smooth on a e8200 + g35(x3500/i965) under ubuntu 8.10 hardy. Olivier Jason Tackaberry wrote: On Sat, 2008-07-26 at 22:18 +0200, Olivier Arsac wrote: I'm wondering why I get these drops... I was under the impression than other people were having better results with slower CPUs... (or were they not outputting to a FullHD display?) My E6600 (2.4Ghz) can handle cornel_m1080p.mov without dropped frames. Yours should very easily as well. [swscaler @ 0x89357b0]SwScaler: using unscaled yuv420p -> rgb32 special converter VO: [gl] 1920x1080 => 1920x1080 BGRA [fs] [zoom] Right, so here's the problem. Don't use the gl VO. You see that first line I quoted, it means that MPlayer is doing yv12 to rgb32 colorspace conversion in software. For 1080p that's a lot of overhead, and it's the reason you're dropping frames. You should use the Xv VO. -vo xv. Cheers, Jason. z - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=""> ___ Freevo-users mailing list Freevo-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freevo-users - This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/___ Freevo-users mailing list Freevo-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freevo-users
Re: [Freevo-users] Help with benchmarking CPUs for HD decoding
Thank you Jason for these very useful advices. Here are my new results. BTW: I've also upgraded the intel xf86 driver to 2.4.0 as the changelog stated numerous speed improvements (and xvmc support... not that useful for this test but always nice to have) Still having frame drops... (less but quite noticeable) A: 119.9 V: 119.9 A-V: 0.009 ct: 0.037 3597/3597 26% 60% 1.2% 42 0 I'm wondering why I get these drops... I was under the impression than other people were having better results with slower CPUs... (or were they not outputting to a FullHD display?) for reference if I play the 1080p version of big buck bunny I'm getting no frame drop... http://wcdata2.sun.com/08D12355/big_buck_bunny_1080p_h264.mov mplayer -fs cornell_m1080p.mov -lavdopts fast:skiploopfilter=nonref:threads=2 MPlayer 1.0rc2-4.2.3 (C) 2000-2007 MPlayer Team CPU: Intel(R) Core(TM)2 Duo CPU E8200 @ 2.66GHz (Family: 6, Model: 23, Stepping: 6) CPUflags: MMX: 1 MMX2: 1 3DNow: 0 3DNow2: 0 SSE: 1 SSE2: 1 Compiled with runtime CPU detection. mplayer: could not connect to socket mplayer: No such file or directory Failed to open LIRC support. You will not be able to use your remote control. Playing cornell_m1080p.mov. ISO: File Type Major Brand: Original QuickTime Quicktime/MOV file format detected. [mov] Audio stream found, -aid 0 [mov] Video stream found, -vid 1 VIDEO: [avc1] 1920x1080 24bpp 15.385 fps 0.0 kbps ( 0.0 kbyte/s) Clip info: copyright: Copyright 2005 Cornell Lab of Ornithology name: Cornell Lab of Ornithology comments: Produced by the Macaulay Library [gl] using extended formats. Use -vo gl:nomanyfmts if playback fails. xscreensaver_disable: Could not find XScreenSaver window. == Opening video decoder: [ffmpeg] FFmpeg's libavcodec codec family Selected video codec: [ffh264] vfm: ffmpeg (FFmpeg H.264) == == Forced audio codec: mad Opening audio decoder: [faad] AAC (MPEG2/4 Advanced Audio Coding) AUDIO: 44100 Hz, 2 ch, s16le, 128.0 kbit/9.07% (ratio: 16000->176400) Selected audio codec: [faad] afm: faad (FAAD AAC (MPEG-2/MPEG-4 Audio) decoder) == AO: [alsa] 48000Hz 2ch s16le (2 bytes per sample) Starting playback... VDec: vo config request - 1920 x 1080 (preferred colorspace: Planar YV12) Could not find matching colorspace - retrying with -vf scale... Opening video filter: [scale] VDec: using Planar YV12 as output csp (no 0) Movie-Aspect is undefined - no prescaling applied. [swscaler @ 0x89357b0]SwScaler: using unscaled yuv420p -> rgb32 special converter VO: [gl] 1920x1080 => 1920x1080 BGRA [fs] [zoom] ... A: 119.9 V: 119.9 A-V: 0.009 ct: 0.037 3597/3597 26% 60% 1.2% 42 0 Jason Tackaberry wrote: Hi Olivier, On Tue, 2008-07-22 at 17:14 +0200, olivier arsac wrote: I have a problem though... I'm unable to get a run without dropping frames even on a 3.2GHz core 2 duo (penryn e8200 OC) If I disable sound output the test run smoothly... something fishy here. If you disable sound, MPlayer does not need to maintain sync, and so it decides not to drop any frames. frame drop test: $mplayer -fs cornell_m1080p.mov -lavdopts fast:skiploopfilter=nonref The obvious thing here is that you're missing threads=2. You want -lavdopts fast:threads=2, and on that processor you can probably get away with leaving out skiploopfilter=nonref. That cornell video uses slices, so decoding can benefit from multiple threads. (Currently ffmpeg's h264 decoder parallelizes only at the slice level. Frame parallelization is coming.) Most (perhaps all, at least all I've seen) Bluray and HD-DVD video uses slices as well. You can check if the content you're viewing has slices by passing -lavdopts debug=1 -v. benchmark: real0m34.380s user0m47.687s sys0m0.304s seems to have plenty of room for real time playback... Yes, but the number of cycles needed to decode any given frame will vary. For example, in that video, the sequence around 70s is quite demanding and will drop frames (at least on my E6600) without specifying threads=2. For single-sliced content, you're more likely to run into trouble. I've done a few test transcodes of 1080p Bluray content with x264 and at crf=22 (with AQ) the bitrate was low enough to still manage without dropped frames on a single core. In theory with a higher bitrate single-sliced h264 content you could start dropping frames. There we have two options: try skiploopfilter=all (much more visible quality loss), or wait a few months (hopefully just that!) for frame-level parallelization with h264 decoding. Cheers, Jason. ---
Re: [Freevo-users] Help with benchmarking CPUs for HD decoding
Here are my benchmarcks for the cornell file played on a fullHD TV. (ref: http://www.mail-archive.com/freevo-users@lists.sourceforge.net/msg14734.html) I have a problem though... I'm unable to get a run without dropping frames even on a 3.2GHz core 2 duo (penryn e8200 OC) If I disable sound output the test run smoothly... something fishy here. Could some of you take a look and give me some advice? *e8200 @ 2.6GHz (8x333MHz) * *frame drop test: *$mplayer -fs cornell_m1080p.mov -lavdopts fast:skiploopfilter=nonref A: 119.9 V: 119.9 A-V: 0.002 ct: 0.037 3597/3597 34% 60% 1.1% 223 0 *benchmarck:* $ time mplayer cornell_m1080p.mov -nosound -vo null -benchmark -lavdopts fast:skiploopfilter=nonref MPlayer 1.0rc2-4.2.3 (C) 2000-2007 MPlayer Team CPU: Intel(R) Core(TM)2 Duo CPU E8200 @ 2.66GHz (Family: 6, Model: 23, Stepping: 6) CPUflags: MMX: 1 MMX2: 1 3DNow: 0 3DNow2: 0 SSE: 1 SSE2: 1 Compiled with runtime CPU detection. mplayer: could not connect to socket mplayer: No such file or directory Failed to open LIRC support. You will not be able to use your remote control. Playing cornell_m1080p.mov. ISO: File Type Major Brand: Original QuickTime Quicktime/MOV file format detected. [mov] Audio stream found, -aid 0 [mov] Video stream found, -vid 1 VIDEO: [avc1] 1920x1080 24bpp 15.385 fps0.0 kbps ( 0.0 kbyte/s) Clip info: copyright: Copyright 2005 Cornell Lab of Ornithology name: Cornell Lab of Ornithology comments: Produced by the Macaulay Library == Opening video decoder: [ffmpeg] FFmpeg's libavcodec codec family Selected video codec: [ffh264] vfm: ffmpeg (FFmpeg H.264) == Audio: no sound Starting playback... VDec: vo config request - 1920 x 1080 (preferred colorspace: Planar YV12) VDec: using Planar YV12 as output csp (no 0) Movie-Aspect is undefined - no prescaling applied. VO: [null] 1920x1080 => 1920x1080 Planar YV12 [zoom] V: 119.9 3597/3597 32% 0% 0.0% 0 0 BENCHMARKs: VC: 39.411s VO: 0.007s A: 0.000s Sys: 0.329s = 39.747s BENCHMARK%: VC: 99.1551% VO: 0.0170% A: 0.% Sys: 0.8279% = 100.% Exiting... (End of file) real0m39.774s user0m56.772s sys0m0.296s (the benchmarks runs are very consistent... less than 1s of difference across 5 runs) *e8200 @ 3.2GHz (8*400MHz)* *benchmark:* real0m34.380s user0m47.687s sys0m0.304s seems to have plenty of room for real time playback... *drop frames:* - with file on hardrive even with the faster CPU still a lot of frame drops (very noticeable on the HDTV each time there is a scrolling) A: 119.9 V: 119.9 A-V: 0.003 ct: 0.035 3597/3597 31% 59% 1.2% 85 0 - with file in /dev/shm/ (no hard drive activity) I get the same droprate average than on hard drive so the I/O are probably not in cause. A: 119.9 V: 119.9 A-V: 0.007 ct: 0.037 3597/3597 31% 59% 1.2% 84 0 with -nosound no frame drop at all? is it just a side-effect of the small amount of CPU that is not used to process sound or a real problem with the sound subsystem / process scheduler? (BTW: with sound activated there is no noticeable sound glitch or stutter.) V: 119.9 3597/3597 31% 60% 0.0% 0 0 with -ao alsa alsa, oss... no big difference A: 119.9 V: 119.9 A-V: 0.012 ct: 0.036 3597/3597 31% 59% 1.2% 76 0 with snice +20 mplayer same range of dropped frames... not the scheduler fault? dstat gives me 10..45% usr+sys CPU load (2 cores so the 45+ may indicate spikes that go other the 50% mark?) I've watched for CPU throttling but the freq is stable a 3.2GHz (I've even forced the performance governor to be sure there's no CPU downclock during the tests) CPU/sys temps are fine (around 55C according sensors) *Various hardware related informations:* Asus p5e-vm hdmi (p35 + g35 (i965)) [EMAIL PROTECTED] (8x400, nothing else changed -> rock solid and very conservative OC perfect for a silent HTPC) 2G DDR2 1066 (pc8000) (mem86 test OK) samsung F1 750g + 500g (ahci mode enabled in bios) *Various software related informations: *Using an uptodate Ubuntu 8.04 (hardy heron) $ cat /proc/version Linux version 2.6.24-19-generic ([EMAIL PROTECTED]) (gcc version 4.2.3 (Ubuntu 4.2.3-2ubuntu7)) #1 SMP Fri Jul 11 23:41:49 UTC 2008 X.Org X Server 1.4.0.90 Intel driver 2.2.0 MPlayer 1.0rc2-4.2.3 (mplayer vo is gl) alsa-base 1.0.16-0ub $alsamixer: Card: HDA Intel │ Chip: Generic 1095 SI HDMI $ cat /etc/drirc (vblank_mode = 3 to avoid tearing) X is running in full HD (1920x1080) $ xdriinfo Screen 0: i965 $ cat /proc/version Linux version 2.6.24-19-generic ([EMAIL PROTECTED]) (gcc version 4.2.3 (Ubuntu 4.2.3-2ubuntu7)) #1 SMP Fri Jul 11 23:41:49 UTC 2008 $ glxinfo ... direct rendering: Yes ... sound output is done on the spdif (digital coaxial output) using cornell test file I get: Forced audio codec: mad Openin