[HN Gopher] Intel becomes the first to offer full AV1 support wi...
       ___________________________________________________________________
        
       Intel becomes the first to offer full AV1 support with its new Arc
       GPUs
        
       Author : clouddrover
       Score  : 75 points
       Date   : 2022-04-03 13:29 UTC (9 hours ago)
        
 (HTM) web link (www.neowin.net)
 (TXT) w3m dump (www.neowin.net)
        
       | zionic wrote:
       | Finally! Given the large overall percentage of web bandwidth
       | that's video this will really help.
       | 
       | Hopefully Apple pushes this over HEVC/VVC garbage in their next
       | release.
        
         | GeekyBear wrote:
         | Google's own devices don't have hardware support for AV1
         | playback yet.
         | 
         | It's probably a bit early to complain about Apple.
        
           | EddieRingle wrote:
           | > Google's own devices don't have hardware support for AV1
           | playback yet.
           | 
           | The Tensor SoC in the Pixel 6 supports AV1 decoding:
           | 
           | https://www.anandtech.com/show/17032/tensor-soc-
           | performance-...
        
           | ChuckNorris89 wrote:
           | > _It 's probably a bit early to complain about Apple._
           | 
           | Why? Apple mints new silicon by its own design every year
           | like swiss clock work. Google on the other hand, depends on
           | the likes of Quallcomm, Mediatek and Samsung for the silicon,
           | whenever they decide to add AV1 IP blocks, leading to a
           | chicken and egg situation.
           | 
           | Google could probably fore their hand by switching to AV1
           | before HW support arrives, but then Youtube UX will suck on
           | all smartphones leading to shit battery life and overheating.
           | 
           | Plus, video codec IP blocks are always a hot potato for the
           | SoC vendors as they're tied in various third party licensing
           | and patent schemes they don't like to deal with.
        
             | SllX wrote:
             | Yeah but given the lead times on Apple silicon, I would
             | expect that if they hadn't already made the decision to
             | support AV1, then we're not going to see it for a few
             | years.
             | 
             | Conversely if they made the decision a few years ago, we
             | might start to see support trickle out across their lines
             | starting this year.
        
               | ZeroGravitas wrote:
               | With VP9 Apple added decode to their custom hardware and
               | then didn't tell anyone until they started to support it
               | in software a few years later.
        
             | GeekyBear wrote:
             | > Google could probably fore their hand by switching to AV1
             | before HW support arrives, but then Youtube UX will suck on
             | all smartphones leading to shit battery life and
             | overheating.
             | 
             | Google could also face yet more antitrust action if they
             | continue to use their Youtube monopoly as a weapon against
             | competition as they previously have with Microsoft's
             | Windows Phone and Amazon's Echo Show.
        
               | phkahler wrote:
               | AV1 isnt just Google, the standard if free for anyone to
               | use.
               | 
               | Google may be guilty of some things, but pushing AV1
               | being an antitrust violation is not one of them.
        
               | GeekyBear wrote:
               | Google has certainly used it's Youtube monopoly against
               | Windows Phone and Amazon's Echo Show and Fire TV in the
               | past.
               | 
               | Which makes it unwise to continue to talk about how
               | Google can use their Youtube monopoly again to force it's
               | competitor's hands on anything.
        
         | CharlesW wrote:
         | > _Hopefully Apple pushes this over HEVC /VVC garbage in their
         | next release._
         | 
         | I think AV1 is neat, but calling it HEVC "garbage" seems harsh
         | considering the enormous leap H.265 represents over H.264. Even
         | if AV1 is ~20% more efficient than HEVC, lots of "better"
         | codecs have failed to become mainstream.
         | 
         | Because VVC appears to trounce* AV1, Apple may just skip AV1.
         | (*https://www.winxdvd.com/video-transcoder/h266-vvc-vs-
         | av1.htm....)
         | 
         | > _...this will really help._
         | 
         | An Intel-made laptop GPU is not going to be the thing that
         | makes or breaks AV1.
        
           | mchusma wrote:
           | VVC is great, but AV1 is fast enough for realtime use, while
           | VVC is not. Cisco has had it working for some time.
           | Basically, AV1 to me seems like it's basically a zero
           | tradeoff solution vs older codecs, so is a good "new
           | standard" to base around then think about what is next from
           | here.
           | 
           | https://blog.webex.com/video-conferencing/cisco-leap-
           | frogs-h...
        
             | CharlesW wrote:
             | > _VVC is great, but AV1 is fast enough for realtime use,
             | while VVC is not._
             | 
             | Today, in software. But yes, definitely worth considering!
        
         | whydid wrote:
         | They have hardware acceleration for encoding and decoding HEVC
         | in all their products. Why would they promote a new codec
         | without hardware acceleration?
        
         | miohtama wrote:
         | Apple is a stakeholder in HEVC patents so I would expect them
         | to fight to the bitter end against non-Apple invented
         | technologies.
        
           | garblegarble wrote:
           | Apple is also a member of the alliance making AV1, The
           | Alliance For Open Media
        
             | tux3 wrote:
             | Apple's history of membership in AOM is complicated(tm).
             | 
             | They joined late and became a founding member
             | retroactively, but I wouldn't use that as a very strong
             | signal.
             | 
             | Apple is never the first to support open standards.
        
               | ysleepy wrote:
               | html5 ? caldav/carddav (their spec though)? usb-c?
               | avahi/mdns? they do sometimes get a standards boner, it's
               | just very specific and strategic.
        
               | AshamedCaptain wrote:
               | Ah yes, that Apple that at some point even adopted
               | Jabber. Doesn't really exist anymore...
        
               | posguy wrote:
               | Yet HTML5 isn't a first class citizen on Apple's most
               | popular operating system, nor is USB-C even on Apple's
               | current best selling product.
        
       | rektide wrote:
       | I'm double surprised! I really thought av1 was.gonna take a long
       | time to encode. I wonder if it can do the same as before... can
       | it do 4k60? I dunno what existing GPUs can do tbh.
       | 
       | Also surprised the efficiency is t that wildly much of a leap.
       | 50% over h264, 20% over h265. Maybe this is part of the tradeoff
       | when trying to encode realtime, versus offline & in multiple
       | passes. I feel like for offline encoding there's way bigger gains
       | but I dunno!
        
       | mcraiha wrote:
       | Also software decoding of AV1 is already fast enough, so
       | specially what game streaming people need is the AV1 hardware
       | encoder https://openbenchmarking.org/test/pts/dav1d#results
        
         | baybal2 wrote:
         | > 20% over h265
         | 
         | On what codec settings?. IMO h256 and AV1 are extremely close
         | depending on individual, and still experimental, codec tweaks.
         | 
         | MP3 coding was near garbage at 128kbps, but over the years the
         | codec community managed to make 128kbps significantly better
         | audibly without changing the format
         | 
         | Next gen h256 coders will likely copy whatever non-bitstream
         | optimisation they can from AV1 coders, and they will be a near
         | match again.
        
           | phkahler wrote:
           | >> Next gen h256 coders will likely copy whatever non-
           | bitstream optimisation they can from AV1 coders, and they
           | will be a near match again.
           | 
           | Not only is that speculation, but more importantly they will
           | not be copying the licensing terms of AV1.
        
           | joshcryer wrote:
           | AV1 beats h265 by some 10-15% and it's still improving. h265
           | has had plenty of time to mature. I don't think there's much
           | headroom for h265 and anyway the JVET/MPEG has moved on to
           | VVC, which beats AV1 by some 25-29% and it is still
           | improving. They know they'll lose the short term war and are
           | looking longer term.
           | 
           | AV1 was supposed to be "on par" with h265 but has jumped
           | ahead by a bit and will likely improve for quite awhile until
           | AV2 is worked on. Will it reach VVC levels? Probably not. But
           | it will certainly be better than h265/HVEC.
           | 
           | Source for size comparisons:
           | https://www.cambridge.org/core/journals/apsipa-
           | transactions-...
        
             | kingcharles wrote:
             | VVC honestly astounds me. It makes me wonder how far we can
             | go with video compression.
             | 
             | I look back at MPEG-1 Video CDs and it is unbelievable how
             | incredible a video image you could now produce in the same
             | 600Mb storage.
        
         | karmakaze wrote:
         | > While AMD and Nvidia both offer AV1 decoding with their
         | newest GPUs, neither have support for AV1 encoding.
         | 
         | So this announcement is specifically about encoding.
        
       | gotaquestion wrote:
       | Surprised to see Intel using "ARC" as a trademark when ARC
       | international has been around for ages and bought by Synopsys in
       | 2010, which now offers ARC processors. I thought trademarks in
       | the same domain weren't allowed?
        
         | JaimeThompson wrote:
         | Even stranger is Intel Ark has been their specs repo for a long
         | time. https://ark.intel.com/content/www/us/en/ark.html
        
       | clouddrover wrote:
       | The AV1 section of the Intel Arc Graphics announcement video:
       | 
       | https://youtu.be/q25yaUE4XH8?t=598
        
       | jalino23 wrote:
       | I'm new to codecs coming from frontend webdev mostly JS. is
       | anyone able to point me to some resources on how to know via cli
       | or programmatically to detect if a hardware codec is installed?
       | and if it's currently being utilized by web video player
       | 
       | the two ways I recently learned is only for listing codecs.
       | 
       | for windows is via legacy media player
       | 
       | for linux is via ffmpeg
       | 
       | and I'm still confused how video players talk to codecs and why
       | they are installed separately and how code can interface with
       | them if they are installed seperately
        
       | chews wrote:
       | Built in Puerto Rico fabs if I'm not mistaken
        
         | wmf wrote:
         | No, I don't think TSMC N6 is made in Puerto Rico.
        
       ___________________________________________________________________
       (page generated 2022-04-03 23:02 UTC)