Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Video Encoding > VP9 and AV1

Reply
 
Thread Tools Search this Thread Display Modes
Old 1st October 2018, 16:17   #1041  |  Link
Beelzebubu
Registered User
 
Join Date: Feb 2003
Location: New York, NY (USA)
Posts: 109
Quote:
Originally Posted by benwaggoner View Post
And we know that the best HEVC encoders can outperform the best VP9 encoders by easily 20-30% for real-world scenarios, so a quote assuming VP9 and HEVC are equivalent just invalidates the AV1 comparison.

[..]

libvpx
I'm all for pointing out PR for what it is, but to equate "libvpx" with "best VP9 encoders" is inherently unfair as an argument against an actually-good VP9 encoder.
Beelzebubu is offline   Reply With Quote
Old 1st October 2018, 17:44   #1042  |  Link
LigH
German doom9/Gleitz SuMo
 
LigH's Avatar
 
Join Date: Oct 2001
Location: Germany, rural Altmark
Posts: 6,753
The media-autobuild_suite already allows using libdav1d in ffmpeg; unfortunately, from time to time, other libraries may break it, so consider to exclude what you don't really need...
__________________

New German Gleitz board
MediaFire: x264 | x265 | VPx | AOM | Xvid
LigH is offline   Reply With Quote
Old 1st October 2018, 19:24   #1043  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 4,750
Quote:
Originally Posted by NikosD View Post
I think the same could be said for HEVC too compared to AVC.

I remember the days when x265 was claiming 50% more compression for the same quality compared to x264, but you have to dig too much to find such a case, if any.
I have seen 50% reduction at very low bitrates and UHD resolutions. But for moderately grainy SD/HD content, x264 versus x265 is more like 30%.

Grain/noise parameterization and reconstruction is probably the single biggest next step in encoding performance, since random noise is intrinsically uncompressible.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 1st October 2018, 19:37   #1044  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 4,750
Quote:
Originally Posted by Beelzebubu View Post
I'm all for pointing out PR for what it is, but to equate "libvpx" with "best VP9 encoders" is inherently unfair as an argument against an actually-good VP9 encoder.
I wasn't specifically calling out libvpx here, but I've not been able to get a real-world comparison of anything substantially better in terms of potential quality (not even quality @ perf).

I'm quite curious about what Eve can really do! But I've struggled to find samples for which source is available so I can try a real apples-to-apples. Their website seems to just have frames, not even any actual video examples.

If anyone can point me towards any, I'd really appreciate it, and could probably make some comparative HEVC encodes available.

For example, here's a recent Tears of Steel x265 test I did (1.5 Mbps average 4 Mbps peak, max 5 sec GOP, unlimited encoding time)

https://1drv.ms/v/s!AlvIQZWsyeO-kKplp2EQ8-Q4bCNVZw

It is pretty awesome that we can deliver a pretty good 1080p experience at VideoCD bitrates! 20x more pixels, and I think better per-pixel quality.

I've been doing 1, 1.5, and 2 Mbps ToS encodes using x265, x264, xvid, WMV VC-1, and VC-1 adaptive resolution Smooth Streaming. Hoping to kick off a VP9 set today, but it's been surprisingly hard to get good detailed parameter tuning documentation for libvpx.

I really hope libaom will get something as good as x265.readthedocs.io! That's really the gold standard of encoder documentation to date.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 1st October 2018, 22:43   #1045  |  Link
TD-Linux
Registered User
 
Join Date: Aug 2015
Posts: 34
Quote:
Originally Posted by blurred View Post
"The most important part of the change is that Google has separated the patent license from the copyright license. Now the copyright license on the software is a totally standard three-clause BSD license, which is clearly compatible with the GPL. The patent license, in turn, provides distributors with permission to exercise all the rights, and meet all the conditions, in the GPL, as required by GPLv2 section 7; and those permissions are consistent with the ones provided by the patent grant in GPLv3 section 11. All this means that developers distributing GPL-covered software can take advantage of the patent license without running afoul of the GPL's conditions, whether they're using GPLv2 or GPLv3."

Does it still apply to AV1?
Yes, it does.
TD-Linux is offline   Reply With Quote
Old 2nd October 2018, 00:35   #1046  |  Link
Blue_MiSfit
Derek Prestegard IRL
 
Blue_MiSfit's Avatar
 
Join Date: Nov 2003
Location: Los Angeles
Posts: 5,988
All the improvements in dav1d is fabulous. Up to 50fps for 8 bit 1080p with 4 threads AND NO ASM YET is pretty amazing.
Blue_MiSfit is offline   Reply With Quote
Old 2nd October 2018, 00:37   #1047  |  Link
Mr_Khyron
Member
 
Mr_Khyron's Avatar
 
Join Date: Nov 2002
Posts: 203
Introducing dav1d: a new AV1 decoder

http://www.jbkempf.com/blog/post/2018/Introducing-dav1d
Mr_Khyron is offline   Reply With Quote
Old 2nd October 2018, 07:14   #1048  |  Link
LigH
German doom9/Gleitz SuMo
 
LigH's Avatar
 
Join Date: Oct 2001
Location: Germany, rural Altmark
Posts: 6,753
And wiiaboo + schmidthubert made ffmpeg build again. So build your favourite kind of ffmpeg with dav1d. — Sorry, too early, API may have changed, some exports are not found suddenly.



@benwaggoner: PM?
__________________

New German Gleitz board
MediaFire: x264 | x265 | VPx | AOM | Xvid

Last edited by LigH; 2nd October 2018 at 08:38.
LigH is offline   Reply With Quote
Old 2nd October 2018, 15:14   #1049  |  Link
Adonisds
Registered User
 
Join Date: Sep 2018
Posts: 14
Quote:
Originally Posted by Nintendo Maniac 64 View Post
I just did some performance testing with the 1080p 30fps AV1 encode of the Gus Kenworthy & Tom Wallisch X Games Slopestyle GoPro Preview video in MPC-HC v1.8.1 x64 with its built-in LAVfilters; I originally tried the Halo video but I found the X Games video to be much more demanding (but also much more motion-sick inducing, especially when playing at slower than real-time).

With my 4c/8t Nehalem Xeon x3470 I was only seeing ~25% CPU utilization at maximum even though I was unable to play back the video in real-time (it was somewhere between 16fps and 20fps). Mathematically that should mean that it's only using 2 threads, but disabling SMT and setting my BIOS to only enable 2 cores resulted in noticably worse performance, yet setting the BIOS to enable 3 cores without SMT resulted in the same 16-20fps performance I was originally seeing yet at only ~67% CPU utilization.

At least with the LAVfilters bundled with MPC-HC v1.8.1 x64, it would seem that the AV1 decoder can only utilize 3 cores and no SMT, yet even then the 2 less loaded cores are only hitting around half of their according core's available utilization.


And for reference, the 720p 30fps AV1 encode of that same video played back without a hitch on my Xeon - heck it left enough headroom that I could turn on a bunch of motion interpolation which greatly helped alleviate the motion sickness I got from watching the 1080p AV1 encode playback at sub-20fps frame rates (it's times like this that I thank the devs over at Nintendo for making F-Zero X and F-Zero GX native 60fps games).



Now I'm a bit out-of-the-loop, but I couldn't help but notice that YouTube-DL was using the .MP4 extension for AV1 downloads - is that in fact correct behavior? (and no, I don't mean AVC1, otherwise my PC would have been playing back the videos easy-peasy).
How do you turn on motion interpolation?
Adonisds is offline   Reply With Quote
Old 2nd October 2018, 18:44   #1050  |  Link
Mr_Khyron
Member
 
Mr_Khyron's Avatar
 
Join Date: Nov 2002
Posts: 203
AV1 encoder project from Intel

https://itpeernetwork.intel.com/open...-visual-cloud/
Quote:
Next Gen CODECs

Streaming media is the foundation for visual cloud workloads. Our open source projects will also advance the next generation of CODECs, specifically Scalable Video Technology (SVT) for HEVC and AV1.

Additionally, Intel is contributing to the SVT-HEVC Encoder core to enable high performance, quality, and scalability of HEVC video encoding under a highly permissive BSD and patent license. This HEVC-compliant encoder library core achieves excellent density-quality tradeoffs and is highly optimized for Intel® Xeon® Scalable and Intel® Xeon® D processors.

We are also forming a new SVT-AV1 open source encoder project as an enhancement to the Alliance for Open Media (AOM) to provide a cleaner, easier to use codebase. Community support is critical to open source innovation, and Intel welcomes contributions to the SVT-AV1 project. Register for email updates at 01.org.

As video consumption and generation continues to grow, so will the number of industries that must deliver high-bandwidth, low-latency video at scale. Open source software is fundamental to meeting the demands, and Intel is committed to collaborating with industry leaders to grow the community.
Mr_Khyron is offline   Reply With Quote
Old 2nd October 2018, 23:07   #1051  |  Link
NikosD
Registered User
 
Join Date: Aug 2010
Location: Athens, Greece
Posts: 2,901
Quote:
Originally Posted by hajj_3 View Post
https://www.youtube.com/watch?v=UhIgBdrKyNM - Dav1d: a fast new AV1 decoder
This is a very fast decoder.

Using just C with almost no ASM and no SIMD at all, they managed to be 60% faster than libaom v1.0.0.0 and capable of 1080p decoding of a 4Mbps clip at 50fps using 4 cores.

In the presentation they said that they are expecting the SIMD version to hit a 4x decoding acceleration, meaning 200 fps for 1080p using 4 cores.

That would be probably the biggest step up in software decoding SIMD acceleration.
__________________
Win 10 x64 (19042.572) - Core i5-2400 - Radeon RX 470 (20.10.1)
HEVC decoding benchmarks
H.264 DXVA Benchmarks for all
NikosD is offline   Reply With Quote
Old 3rd October 2018, 08:16   #1052  |  Link
NikosD
Registered User
 
Join Date: Aug 2010
Location: Athens, Greece
Posts: 2,901
If someone could build and provide latest libaom and dav1d decoders in DirectShow filter form, I'm interested in opening a new thread for AV1 decoder's evaluation.

I have various CPUs from Core 2 Duo to Haswell, so I can follow the improvements in ASM, MultiThreading and SIMD acceleration for both AV1 decoders.

BTW, are there any other AV1 decoders ?

But I need them to be registered as DirectShow filters.
__________________
Win 10 x64 (19042.572) - Core i5-2400 - Radeon RX 470 (20.10.1)
HEVC decoding benchmarks
H.264 DXVA Benchmarks for all
NikosD is offline   Reply With Quote
Old 3rd October 2018, 08:23   #1053  |  Link
foxyshadis
ангел смерти
 
foxyshadis's Avatar
 
Join Date: Nov 2004
Location: Lost
Posts: 9,558
Quote:
Originally Posted by benwaggoner View Post
I have seen 50% reduction at very low bitrates and UHD resolutions. But for moderately grainy SD/HD content, x264 versus x265 is more like 30%.

Grain/noise parameterization and reconstruction is probably the single biggest next step in encoding performance, since random noise is intrinsically uncompressible.

I have a feeling that video standards are eventually going to have a similar come-to-jesus moment that graphic card makers faced when they were finally called out for totally ignoring minimum FPS. That led to some pretty big driver changes within a year, but video standards have a much longer lead time....
foxyshadis is offline   Reply With Quote
Old 3rd October 2018, 08:54   #1054  |  Link
NikosD
Registered User
 
Join Date: Aug 2010
Location: Athens, Greece
Posts: 2,901
Quote:
Originally Posted by Nintendo Maniac 64 View Post
...I did however notice that, out of the three cores being utilized on my Xeon x3470, one of them is pretty much fully pegged and the other two are about half utilization - this would then equal the ~25% utilization I'm seeing.

Now normally I would let this all go as simply a case of "not having fast enough single-threaded performance" and be done with it, but the fact that you and others were seeing over 60% utilization on your own 4c/8t CPUs (which implies it was balancing the load across more than 4 threads) makes me thing something still isn't quite right here - I mean, weaker single-threaded performance shouldn't result in fewer threads being utilized, and if anything you'd want it to be the opposite, no?
My preliminary results for LAV x64 0.72.0-15 and MPC-HC v1.8.2 using libaom v1.0.0-552 show me only ~75% CPU utilization of a 2C/4T Haswell Core i3-4170@3.7GHz, meaning that the current status of libaom decoder is an up to 3 threads SMT implementation.

The above situation leads to a ceiling in performance of course, meaning that a lot of 1080p AV1 samples out there, are simply not real-time decoded by libaom.

Probably even the no SIMD version of dav1d as it is right now, due to the 4 threads implementation IIRC, could do it better.

But generally speaking, "we need a bigger boat"
__________________
Win 10 x64 (19042.572) - Core i5-2400 - Radeon RX 470 (20.10.1)
HEVC decoding benchmarks
H.264 DXVA Benchmarks for all
NikosD is offline   Reply With Quote
Old 3rd October 2018, 15:19   #1055  |  Link
clsid
*****
 
Join Date: Feb 2005
Posts: 5,642
The number of threads for libaom depends on the number of tiles in the video.

Newer libaom build is already 10-15% faster. You should wait with tests until dav1d gets into FFmpeg/LAV. Then libaom as decoder will become obsolete.
clsid is offline   Reply With Quote
Old 3rd October 2018, 20:03   #1056  |  Link
NikosD
Registered User
 
Join Date: Aug 2010
Location: Athens, Greece
Posts: 2,901
Quote:
Originally Posted by clsid View Post
The number of threads for libaom depends on the number of tiles in the video.
The number of tiles is a property of the video created by the encoder or is it something that is been created by the decoder ?
__________________
Win 10 x64 (19042.572) - Core i5-2400 - Radeon RX 470 (20.10.1)
HEVC decoding benchmarks
H.264 DXVA Benchmarks for all
NikosD is offline   Reply With Quote
Old 3rd October 2018, 20:08   #1057  |  Link
LigH
German doom9/Gleitz SuMo
 
LigH's Avatar
 
Join Date: Oct 2001
Location: Germany, rural Altmark
Posts: 6,753
The encoder decides that, usually based on the resolution of the video, especially the height, if it was not specified expliticly as parameter.
__________________

New German Gleitz board
MediaFire: x264 | x265 | VPx | AOM | Xvid
LigH is offline   Reply With Quote
Old 3rd October 2018, 21:21   #1058  |  Link
NikosD
Registered User
 
Join Date: Aug 2010
Location: Athens, Greece
Posts: 2,901
By watching the presentation of dav1d, he gave me the impression that a tile-based decoder decides the number, the scheme, the orientation etc of the tiles, but without saying too many details in the presentation.

Wrong impression obviously.
__________________
Win 10 x64 (19042.572) - Core i5-2400 - Radeon RX 470 (20.10.1)
HEVC decoding benchmarks
H.264 DXVA Benchmarks for all
NikosD is offline   Reply With Quote
Old 3rd October 2018, 22:37   #1059  |  Link
Beelzebubu
Registered User
 
Join Date: Feb 2003
Location: New York, NY (USA)
Posts: 109
Sorry if that created confusion; tiles (and thus the amount of tile threads a decoder can use) are set by encoder; frame threads is selectable by decoder without requiring special encoder settings.
Beelzebubu is offline   Reply With Quote
Old 3rd October 2018, 22:48   #1060  |  Link
user1085
Registered User
 
Join Date: Apr 2018
Posts: 22
Quote:
Originally Posted by Beelzebubu View Post
Sorry if that created confusion; tiles (and thus the amount of tile threads a decoder can use) are set by encoder; frame threads is selectable by decoder without requiring special encoder settings.
For comparison with other decoders for H.265, H.264, do we know how many threads they use?
user1085 is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 16:31.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.