VP9 encoder is 20x times slower than VP8 !!!

[Alexey Eromenko]

As you suggested, I did convert my WebM to raw video, using the
following command:

ffmpeg -i Skyrim.webm -f yuv4mpegpipe - > Skyrim.y4m

(RAW video itself plays fine on VLC player)

VP 9 encoder (after 15 minutes of run-time):

root@deb7vm:~/libvpx# ./vpxenc -v --target-bitrate=600 -o
skyrim-vp9-600kbps.webm --codec=vp9
/media/sf_C_DRIVE/Multimedia/Clips-WebM/Skyrim.y4m
Codec: WebM Project VP9 Encoder v1.2.0-1610-g1e8b3da
Source file: /media/sf_C_DRIVE/Multimedia/Clips-WebM/Skyrim.y4m Format: YV12
Destination file: skyrim-vp9-600kbps.webm
...
Pass 1/1 frame 29/28 22130B 1094158 ms 37729.61 ms/f [ETA 19:08:21]

VP 8 encoder (after 15 minutes of run-time):

root@deb7vm:~/libvpx# ./vpxenc -v --target-bitrate=600 -o
skyrim-vp8-600kbps.webm --codec=vp8
/media/sf_C_DRIVE/Multimedia/Clips-WebM/Skyrim.y4m
Codec: WebM Project VP8 Encoder v1.2.0-1610-g1e8b3da
Source file: /media/sf_C_DRIVE/Multimedia/Clips-WebM/Skyrim.y4m Format: YV12
Destination file: skyrim-vp8-600kbps.webm
...
Pass 1/1 frame 671/670 1477650B 1209219 ms 1802.12 ms/f [ETA 0:47:34]

Same hardware. (VirtualBox 4.2.x on Core i7 2600 CPU)

Do you see the problem here ?
One coder (VP8) has encoded over 600 frames, the other (VP9) only 29
frames. (in 15 minutes).
2 frames per minute ? yay !

Video is HD 720p.
Why VP9 encodes so slow ? Is it computing on a wooden abacus ?

--
-Alexey Eromenko "Technologov"

[Alexey Eromenko]

One practical implication, is that it precludes me from doing a
real-world testing of VP9. (I have lots of videos to play with...
ranging from ugly low-bitrate 240p up to 1080p Full HD blue-ray)

Issue 527: vp9 encoder is slow
--
-Alexey Eromenko "Technologov"

[Ronald S. Bultje]

Hi Alexey,

speed isn't currently our focus with VP9. Our current focus is to
create a video codec that is awesome. Once the spec is finished, we'll
focus more on the speed aspect of it.

Ronald

> --
> You received this message because you are subscribed to the Google Groups "Codec Developers" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to codec-devel...@webmproject.org.
> To post to this group, send email to codec...@webmproject.org.
> Visit this group at http://groups.google.com/a/webmproject.org/group/codec-devel/?hl=en.
> For more options, visit https://groups.google.com/a/webmproject.org/groups/opt_out.
>
>

[Alexey Eromenko]

On Tue, Mar 12, 2013 at 12:51 PM, Ronald S. Bultje <rsbu...@gmail.com> wrote:
> Hi Alexey,
>
> speed isn't currently our focus with VP9. Our current focus is to
> create a video codec that is awesome. Once the spec is finished, we'll
> focus more on the speed aspect of it.

As I said, it creates a very practical problem for BETA testers. For
example, if a 5-minute video clips takes 20 hours to encode, there is
no way for me to help you with testing. Let alone full-length 2-hour
movies...

If the performance penalty could be reduced to become "manageable",
like 2x performance hit, I still could be able to help you.

--
-Alexey Eromenko "Technologov"

[Pjotrs Ivanovs]

Hi, there exist some special test clips, our you can make your own, they short and untouched by compression

i suggest xiph.org test media or/and hevc test media...

//Let alone full-length 2-hour movies...

the is no point to use full lengs movie for beta testing

//Why VP9 encodes so slow ? Is it computing on a wooden abacus ?

i think hi doing Motion Estimation, as there new motion modes ( compound prediction ) and in some configurations more refefernce frames, that is logical that amount of work to be done by encoder increased...

[jcjok...@gmail.com]

What is a good VP 9 encoder?