> doing such claims, please back them up with actual numbers on
> important codecs.
>
> - Hendrik
Benchmarking: perf stat -r 30
CPU: x86_64, 64-bit (Intel Core i3-2120, 3.30GHz)
***seconds time elapsed***
***samples info***
URL or other info
get_bits.h bitstream.h
duration bitrate
HEVC 220.21 (+-0.22%) 220.13 (+-0.24%) - no speed loss 0:12:14
200 kb/s yuv420p, 1920x1080 Divx encode of Tears of Steel
H264 345.98 (+-0.11%) 344.72 (+-0.21%) - no speed loss n/a
200 kb/s yuv420p, 2048x858 looped Gravity Trailer
OPUS 77.71 (+-0.09%) 76.85 (+-0.07%) - 0.7 % speed up 5:36:40
1536 kb/s 48000 Hz
https://www.youtube.com/watch?v=BVp7CTngR8o
AAC 18.76 (+-0.15%) 18.27 (+-0.15%) - 2.7 % speed up
5:36:40 127 kb/s 44100 Hz
https://www.youtube.com/watch?v=BVp7CTngR8o
DCA 18.15 (+-0.08%) 18.06 (+-0.04%) - 0.5 % speed up n/a
1411 kb/s 44100 Hz looped diatonis.dts
***decoder .o size***
get_bits.h bitstream.h
hevc.o 45160 47536
h264_mb.0 137872 139584
opusdec.o 10424 11784
aacdec.o 47608 61464
dcadec.o 42696 53040
CPU: i.MX6@1GHz
URL: https://www.youtube.com/watch?v=9TkHpvaO09c, h264, Duration:
0:04:10, bitrate: 200 kb/s
get_bits.h:
257707.278166 task-clock (msec) # 1.000 CPUs utilized
( +- 0.06% )
660 context-switches # 0.003 K/sec
( +- 16.84% )
13 cpu-migrations # 0.000 K/sec
( +- 2.78% )
7419 page-faults # 0.029 K/sec
256667795435 cycles # 0.996 GHz
( +- 0.06% )
bitstream.h:
258715.049589 task-clock (msec) # 1.000 CPUs utilized
( +- 0.03% )
488 context-switches # 0.002 K/sec
( +- 12.34% )
13 cpu-migrations # 0.000 K/sec
( +- 2.80% )
7409 page-faults # 0.029 K/sec
257672602635 cycles # 0.996 GHz
( +- 0.03% )
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
