Jump to content

[Updated] Apple  WWDC 2022 - What was announced and stuff

Lightwreather
43 minutes ago, hishnash said:

The AV1 codec is open standard but there are no AV1 open hardware decoder/encoders all of that will be a mix of nasty patents. For someone making hardware it might well be much harder to deal with (and much more costly) than HEVC. Even if you design your own you need to ensure you don't accidentally infringe on someone else's implementation. 

Now you're just being ridiculous. 

1) There is a reference decoder accessible to AOMedia members.

2) The patents surrounding hardware accelerated decoding for AV1 are free to use.

3) Are you seriously suggesting that Samsung, Google, AMD, Intel, Nvidia, MediaTek, Allegro, Rockchip, Realtek, Broadcom and Amlogic all managed to implement their own decoders for AV1, but Apple is too scared to infringe some patent? Meanwhile they are designing their own decoding blocks for things like HEVC and H.264 just fine?

4) You do not need to worry about infringing on someone else's design with AV1 if you are a member of AOMedia, which Apple is. Apple could straight up copy the exact design made by Intel, or Nvidia, or AMD (if Apple had access to the design documentation for their implementations), and those companies would not be allowed to make a noise about it. That is part of the agreement that Apple has signed along with everyone else in AOMedia. At least not because of patents surrounding AV1.

 

You are just making up imaginary issues.

 

 

43 minutes ago, hishnash said:

They would need to do the same for any hardware decoder patents they licensed fro AV1.

No they wouldn't. There are no licensing fees for AV1 hardware.

 

 

43 minutes ago, hishnash said:

AV1 is free but only in that anyone cant implement the spec but the spec does not include any suggestion of how to implement it in a power optimal way in modern silicon

This is wrong. AOMedia has reference IP for decoders that members are free to use if they want. Now, you can say what you want about the reference implementation, about how it might not be the most optimized thing in the world, but with all the resources Apple has it would be fairly trivial for them to make their own, and no they do not have to worry about patents. AV1 is heavily based on VP9 which Apple already supports in hardware. On top of that, the new additions to the codec was designed with hardware implementations in mind, with input from the likes of Nvidia and AMD.

 

 

43 minutes ago, hishnash said:

the spec just includes a sub optimal software version as an example

1) I don't think it is "sub optimal".

2) How does this matter exactly? Speaking of reference software, have you tried using the reference software for HEVC? It is beyond usable. It makes AOMedia's reference software look like a miracle. 

3) AOMedia has reference hardware as well.

 

43 minutes ago, hishnash said:

even implementing a software solution of your own of AV1 could run the risk of breaching someone else patents with respect to how that is implemented

No it doesn't. Stop making shit up.

There is some bogus claim from Sisvel about having patents regarding AV1 but that is it. Meanwhile, HEVC requires multiple licenses from multiple patent pools, on top of multiple patents from individual companies, many of which have said they don't even have a price yet, just that they can sue someone for using HEVC if they so desire.

This argument is as ridiculous as saying "I won't eat that Haribo gummy bear because I don't know what it contains. I will instead eat this bag of random pills I found at the hospital because at least I know that some of these pills are aspirin, and I have a headache right now".

You are arguing that a much bigger but partially known risk is better than a minor unknown risk. It makes no sense.

 

43 minutes ago, hishnash said:

there is no license AV1 can create that means others cant patent was to decode it that are not included in the academic examples they provide. 

Yes there is. Stop making shit up.

 

 

Why are you making up so many lies about AV1? It is very clear that you have not made even basic research about it, yet you seem very determined to downplay it and come up with excuses for why Apple doesn't implement it, all of which are based on completely and utterly wrong assumptions.

 

 

Your arguments seem to stem from a few incorrect assumptions.

Assumption number 1: That companies have to design their own hardware for AV1 decoding.

This is incorrect. AOMedia has designed reference implementations that members are free to use if they so desire. It might not be the best thing in the world, but members are also free to create their own.

 

Assumption number 2: That there are a bunch of patents and licenses surrounding AV1 that you need to avoid.

This is incorrect because of the joint patent licensing agreement. All members of AOMedia has agreed to the alliance for open media patent license. This license means that if you use AV1 in any way, you must also allow everyone else to use your AV1-related patents, however they want, in perpetuity, worldwide. The moment someone agrees to this, they sign away their rights to sue someone else for anything AV1 related. This includes Apple, and Apple has already signed this. And yes, this includes hardware. The license makes no distinction between software and hardware.

I highly recommend you read the license agreement. It is very short and easy to read, compared to other licensing agreements. The end result is that nobody has to fear working with AV1. It is by design made to be easy and safe to work with when it comes to implementing it in a product (software or hardware). This is by design to help adoption of it.

Link to comment
Share on other sites

Link to post
Share on other sites

5 hours ago, hishnash said:

nd reset the firmware in the entier system even if a firmware upgrade fails mid way through the existing USB-C port on the device using software you can download from apple

Yeah, but when you have only ONE Apple computer its next to impossible. Thats the issue, this is the first and only Mac I own. No other people I know own a Mac. 

I just want to sit back and watch the world burn. 

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, LAwLz said:

You are arguing that a much bigger but partially known risk is better than a minor unknown risk. It makes no sense.

 

No i'm saying apple cant afford to not have HEVC so adding AV1 is not a comparison to the risk of HEVC it is a comparison to not having AV1. Not just risk but also die area etc.  

Apple cant afford to not have HEVC their entire ecosystem depends on it if they were to ship a product without it then many many users would find browsing there photo lib to be massively energy expensive and for the professional users many modern cameras record and H265 4:2:2 or higher quality doing that on the cpu or gpu would be nasty. 
 

2 hours ago, LAwLz said:

This includes Apple, and Apple has already signed this. And yes, this includes hardware. The license makes no distinction between software and hardware.

You might have highlighted the main risk for apple, if they built there own decoders/encoders they would be forced to share them and according to you share everything about them, if apple were to build there own decoders these would build ontop of existing IP they have in there other decoders, so you saying supporting AV1 would cost apple large parts of their silicon IP?

Apple current HEVC decoders for example include little micro ARM cores using apples own arm arc IP, so that would then also be forced to be usable by any member of that group?

That is why I don't think that IP sharing deal is what you think it is, there is no way Nvidia, AMD, apple etc would risk all there private IP being shared with everyone else due to it being somehow related to AV1 decoding/encoding. I expect the IP sharing only covers patents that explicitly name AV1, anything else including all the patents that actually cover the management of the decoder, the steaming of data to it and from it, the decoding of the data stream in a low power way etc will have been written to not mention AV1 at all. 
 

Link to comment
Share on other sites

Link to post
Share on other sites

Updated Original Post… and that was lot of work

"A high ideal missed by a little, is far better than low ideal that is achievable, yet less effective"

 

If you think I'm wrong, correct me. If I've offended you in some way tell me what it is and how I can correct it. I want to learn, and along the way one can make mistakes; Being wrong helps you learn what's right.

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, hishnash said:

No i'm saying apple cant afford to not have HEVC so adding AV1 is not a comparison to the risk of HEVC it is a comparison to not having AV1. Not just risk but also die area etc.  

Apple cant afford to not have HEVC their entire ecosystem depends on it if they were to ship a product without it then many many users would find browsing there photo lib to be massively energy expensive and for the professional users many modern cameras record and H265 4:2:2 or higher quality doing that on the cpu or gpu would be nasty. 

What?

They could keep HEVC support and then add AV1 in addition. Nobody has suggested they should remove HEVC.

Again, pretty much every single GPU vendor on the planet is already doing this (from small ones like Rockchip to big ones like Intel). It is just Apple (and Qualcomm) dragging their feet. My guess is that Apple will add it sooner or later, but for some reason they seem very against open standards in general. Hell, their SoCs has support for VP9 and yet they don't really expose it. They had it implemented in hardware for years before they finally started adding the software to use it, for some reason.

 

 

The die area is not a real argument either. These chips have 20 billion transistors in them for crying out loud. Adding AV1 support is not even a drop in the bucket.

 

 

 

 

1 hour ago, hishnash said:

You might have highlighted the main risk for apple, if they built there own decoders/encoders they would be forced to share them and according to you share everything about them, if apple were to build there own decoders these would build ontop of existing IP they have in there other decoders, so you saying supporting AV1 would cost apple large parts of their silicon IP?

There is no risk...

Again, you need to get it out of your head that there is some risk with implementing AV1. There is no risk. 

 

They are not forced to share their encoder or decoder logics. They are forced to add their patents revolving AV1 to the patent pool. But the thing is that APPLE HAS ALREADY DONE THIS. Apple has already "paid the price" to be part of AOMedia, but they are not reaping any of the rewards yet.

 

The patent agreement does not include needing to share schematics for your hardware. You only need to share the patents for the AV1 parts. I doubt Apple would even patent the decoding block anyway, so they have nothing to worry about. You usually don't patent the design of your logics. Besides, since the functions are already defined in the AV1 specifications, any attempt to patent the hardware implementation would probably fall under prior art ruling, making the patent invalid.

 

Let me repeat once again, there is no risk with implementing AV1. There are no fees. There are no potential to be sued. The only "risk" is that you have to agree to not sue someone else for AV1, and since Apple is already a member they have already agreed to that part.

 

 

1 hour ago, hishnash said:

Apple current HEVC decoders for example include little micro ARM cores using apples own arm arc IP, so that would then also be forced to be usable by any member of that group?

1) Source on this claim? I don't see why their HEVC decoder would include a "micro ARM core". These things are implemented in specialized ASICs.

2) No, other members would not have access to that even if that was the case, unless Apple for some reason decided to patent the specific design of it. But if it is true that it is an ARM core then they can't patent it anyway.

They might try to copyright the design, but since the design is based on the same specification that everyone else also follows, and the spec is written by a third party, I doubt they would get very far trying to claim copyright. It would be like Apple trying to copyright Intel's socket just because they happen to design the motherboard around the socket.

 

 

The reason, I think, that Apple are dragging their feet, has nothing to do with some "risk" of being sued or whatever, since that risk does not exist. I think the real reason is that Apple doesn't care right now. They have already invested heavily into HEVC, with things like HEIC, so they don't really want to change even if something better exists. That coupled with the fact that, as I said before, they kind of hate open standards for some reason, they don't really want to support it. They implemented VP9 in hardware but it seems like they only allows some apps to use it (like Youtube).

Link to comment
Share on other sites

Link to post
Share on other sites

26 minutes ago, J-from-Nucleon said:

Updated Original Post… and that was lot of work

Not sure if you noticed or not, but the Apple logo in the thread title doesn't seem to show up on Windows and Android (at least for me). Here's what it looks like on Windows 10:

1149412244_WWDCThread.PNG.d6b80ba9c503cf92fbe3e02107d6ef51.PNG

Phobos: AMD Ryzen 7 2700, 16GB 3000MHz DDR4, ASRock B450 Steel Legend, 8GB Nvidia GeForce RTX 2070, 2GB Nvidia GeForce GT 1030, 1TB Samsung SSD 980, 450W Corsair CXM, Corsair Carbide 175R, Windows 10 Pro

 

Polaris: Intel Xeon E5-2697 v2, 32GB 1600MHz DDR3, ASRock X79 Extreme6, 12GB Nvidia GeForce RTX 3080, 6GB Nvidia GeForce GTX 1660 Ti, 1TB Crucial MX500, 750W Corsair RM750, Antec SX635, Windows 10 Pro

 

Pluto: Intel Core i7-2600, 32GB 1600MHz DDR3, ASUS P8Z68-V, 4GB XFX AMD Radeon RX 570, 8GB ASUS AMD Radeon RX 570, 1TB Samsung 860 EVO, 3TB Seagate BarraCuda, 750W EVGA BQ, Fractal Design Focus G, Windows 10 Pro for Workstations

 

York (NAS): Intel Core i5-2400, 16GB 1600MHz DDR3, HP Compaq OEM, 240GB Kingston V300 (boot), 3x2TB Seagate BarraCuda, 320W HP PSU, HP Compaq 6200 Pro, TrueNAS CORE (12.0)

Link to comment
Share on other sites

Link to post
Share on other sites

8 minutes ago, BondiBlue said:

Not sure if you noticed or not, but the Apple logo in the thread title doesn't seem to show up on Windows and Android (at least for me). Here's what it looks like on Windows 10:

1149412244_WWDCThread.PNG.d6b80ba9c503cf92fbe3e02107d6ef51.PNG

That I do know,

That symbol is only there by default on apple OSes, you'd need to install a font to get it to display properly, iirc.

"A high ideal missed by a little, is far better than low ideal that is achievable, yet less effective"

 

If you think I'm wrong, correct me. If I've offended you in some way tell me what it is and how I can correct it. I want to learn, and along the way one can make mistakes; Being wrong helps you learn what's right.

Link to comment
Share on other sites

Link to post
Share on other sites

really great to see apple pushing a password manager. while it's unsure how good it is, unless someone has experience with this one?

Do wish we saw more of that, also being pushed as a standard? could be good, hope to see a good side from this and not pushing their own system onto others.

iphone my android car, android phone my iphone car.

Link to comment
Share on other sites

Link to post
Share on other sites

@LAwLz I've been following AV1 development since before the bitstream was frozen in 2019. You are mostly correct but with a few nitpicks:

 

1. The hardware reference implementation is widely considered terrible and almost outright unusable, so much that AV1 has removed it from their website and a completely rethought hardware reference implementation is being considered for AV2. For this reason, AOMedia recommends licensing hardware decoders from third-party IP providers like Chips&Media. 

 

An engineer from Facebook, David Ronca, on the situation:

Quote

"the reference hardware IP that was provided by AOM was not really optimized and the cost to deploy that in hardware would've been excessive. Now, if you're building 4K TVs, your margins are quite high and maybe that cost is not an issue. When you're deploying Android phones for free, or for $50, or whatever, the cost is critical. So the vendors simply made the decision that they would have to re-do this from the ground up in order to make that hardware codec cost efficient in their chip in terms of transistor count. [...] So that's unfortunate, it's a lesson learned, we'll not make that mistake in AV2."

 

image.thumb.png.242b263e1ad4ca0d9eba5c70a845f178.png

 

Also, if you are thinking this is some very pessimistic Facebook employee, David Ronca has worked in video technologies for over 20 years.

 

2. Because of this, there is a licensing fee on AV1 at least on the decoder IP unless Apple wants to engineer their own encoder or decoder. AV1 is more efficient than HEVC, but this is where problem #3 comes in.

 

3. AV1 is a more efficient codec, but its complexity is much greater for both encode and decode. It was news just a few weeks ago that Intel's AV1 encoder had finally reached the point where it was "only" twice as slow as the HEVC x265 encoder, which is already really slow compared to the AVC x264 encoder. 

 

4. Because of this, Apple may have looked at the situation and decided that the reference decoder wasn't an option, and perhaps third-party encoders weren't at the level Apple expected in terms of quality. Also remember many chips have as much as a ~3 year start-to-finish development time, which means that Apple would have been looking at the situation in ~2019-2020, back when the AV1 encoder performance was abominable and AV1's success far from guaranteed (remember VP9, basically only used by YouTube?). 3rd-party AV1 implementations ~2-3 years ago were also probably not fantastic.

 

5. And to top it all off... some people are claiming patents on AV1. Look up Sisvel which is claiming to have a patent pool on the technology and is demanding licensing fees from non-AOM companies. Apple's Legal department would have to sort out whether AV1 is actually patent free or not. If it is not, it's a decoder (or the cost to design one) and Patent Fees and they were making this decision in ~2019-2020. Not very compelling.

 

Link to comment
Share on other sites

Link to post
Share on other sites

For the reasons above, I can hardly blame Apple for being more on the cautious wait-and-see side of the fence. After all... VP9.

 

Also, even if good reference hardware options are now available, that doesn't mean they are without issues. See the Google Pixel Tensor system. It's a modified Exynos 2100 with Google IP added onto it. It actually has two AV1 decode blocks, one designed by Google and one by Samsung, and the one by Samsung is disabled on the original Exynos and the Tensor rumoredly because of problems (a broken hardware implementation that couldn't be fixed in time). 

 

Two things can be true at once. AV1 can be an amazing new codec the industry should support - and its rollout from a technical hardware perspective has been problematic and troublesome for almost everyone.

Link to comment
Share on other sites

Link to post
Share on other sites

20 minutes ago, gjsman said:

1. The hardware reference implementation is widely considered terrible and almost outright unusable, so much that AV1 has removed it from their website and a completely rethought hardware reference implementation is being considered for AV2. For this reason, AOMedia recommends licensing hardware decoders from third-party IP providers like Chips&Media. 

Well, as I said the reference design was not optimal, but please note that Ronca said the cost wouldn't be an issue for higher end devices, and that it would mostly be an issue for <50 dollar phones.

We are talking about pennies even with the "bad" reference design.

 

Also, this is Apple we are talking about. They had a revenue of almost 380 BILLION dollars last year alone. The money they would need to spend to develop and implement an AV1 decoding block in their SoCs is a rounding error, and they will probably have to do it sooner or later anyway.

 

 

24 minutes ago, gjsman said:

3. AV1 is a more efficient codec, but its complexity is much greater for both encode and decode. It was news just a few weeks ago that Intel's AV1 encoder had finally reached the point where it was "only" twice as slow as the HEVC x265 encoder, which is already really slow compared to the AVC x264 encoder. 

1) I never said I wanted them to do encoding. Decoding is enough for now in my mind.

2) You are talking about software encoding. AV1 is designed to be easy to implement in hardware, and a lot of hardware resources can be shared between the various codecs in the media engine block. Just because software encoding is twice as slow does not mean it is twice as difficult to implement in hardware. Far from it.

3) I am pretty sure you are quoting old numbers, because the whole "STV-AV1 is twice as slow as x265" were news almost 2 years ago, not a couple of weeks ago. And the test was done by Ozer which has historically been a bit lacking on the AV1 encoding side to put it mildly.

4) You are comparing software that have had 9 years of fine tuning (x265), vs software that have been in development for like 3 year (SVT-AV1).

 

 

36 minutes ago, gjsman said:

4. Because of this, Apple may have looked at the situation and decided that the reference decoder wasn't an option, and perhaps third-party encoders weren't at the level Apple expected in terms of quality. Also remember many chips have as much as a ~3 year start-to-finish development time, which means that Apple would have been looking at the situation in ~2019-2020, back when the AV1 encoder performance was abominable and AV1's success far from guaranteed (remember VP9, basically only used by YouTube?). 3rd-party AV1 implementations ~2-3 years ago were also probably not fantastic.

 

That might be true.

 

 

36 minutes ago, gjsman said:

5. And to top it all off... some people are claiming patents on AV1. Look up Sisvel which is claiming to have a patent pool on the technology and is demanding licensing fees from non-AOM companies. Apple's Legal department would have to sort out whether AV1 is actually patent free or not. If it is not, it's a decoder (or the cost to design one) and Patent Fees and they were making this decision in ~2019-2020. Not very compelling.

Sisvel are full of shit.

Besides, have you seen what a mess HEVC is? Sisvel claiming to have some patents related to AV1 but not providing any evidence and not going after anyone is kind of a small threat compared to using HEVC. 

And Apple are an AOMedia member so even if Sisvel are going after non-AOMedia members, why does it matter to Apple? They won't be affected.

Link to comment
Share on other sites

Link to post
Share on other sites

15 minutes ago, LAwLz said:

1) I never said I wanted them to do encoding. Decoding is enough for now in my mind.

Apple actually supports VP9 hardware decode (no encode) in M1 without telling anyone. AV1 could be in M2 right now the same way, but I doubt that. Also, it is completely possible M2 has AV1 but without OS enablement, only time will tell.

 

15 minutes ago, LAwLz said:

3) I am pretty sure you are quoting old numbers, because the whole "STV-AV1 is twice as slow as x265" were news almost 2 years ago, not a couple of weeks ago. And the test was done by Ozer which has historically been a bit lacking on the AV1 encoding side to put it mildly.

 

I looked into it and you are correct on this point. I may have gotten confused and thought it was twice as slow as x265 when it was actually x264 but competitive with x265, or something like that. My bad.

 

15 minutes ago, LAwLz said:

And Apple are an AOMedia member so even if Sisvel are going after non-AOMedia members, why does it matter to Apple?

It matters for a few reasons.

1. Sisvel has patents from non-AOM members which may be legitimate. Because the patent owners are not in AOM, they could be used for litigation. 

2. Apple operates more internationally than most companies. This could end up like MPEG-2, which has 2 patents only valid in Malaysia but MPEG-LA is still extracting royalties on DVD players for them to this day. If Sisvel has even one valid patent, it needs a royalty, possibly on all devices regardless of country of sale (again like MPEG LA's continued patent fees for MPEG-2 even in the US for those Malaysian patents).

3. As for how much money Apple could be on the hook for, remember Apple in 2019 barely managed to vacate a $308 million judgement against themselves for infringing one patent by a patent troll (PMC) over DRM technologies. 

Link to comment
Share on other sites

Link to post
Share on other sites

(some people are really passionate and knowledgeable about AV1 around here…I would add to always take into consideration how far in the past all of these products are planned and eventually freezed/finalized)

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, gjsman said:

 

 

4. Because of this, Apple may have looked at the situation and decided that the reference decoder wasn't an option, and perhaps third-party encoders weren't at the level Apple expected in terms of quality. Also remember many chips have as much as a ~3 year start-to-finish development time, which means that Apple would have been looking at the situation in ~2019-2020, back when the AV1 encoder performance was abominable and AV1's success far from guaranteed (remember VP9, basically only used by YouTube?). 3rd-party AV1 implementations ~2-3 years ago were also probably not fantastic.

 

 

Honestly, there is no excuse to not have a AV1 DECODER by now, but it might be too expensive in chip real estate to have the AV1 encoder on there without sacrificing something. Which is why the Intel and NVIDIA decoders showed up, and only the Intel chip has the encoder. That said Intel's encoders have generally been a mix of software and hardware, and the AV1 support came at the expense of the VP8 support. Not that I think the die space is comparable.

 

That said, by the time AV1 encoder support is available in everything, the H265 patents will all expire and we can move on from H264 and recycle that die space.

Link to comment
Share on other sites

Link to post
Share on other sites

6 minutes ago, Kisai said:

That said, by the time AV1 encoder support is available in everything, the H265 patents will all expire and we can move on from H264 and recycle that die space.

Oh heavens no, hopefully not that late. H.265 still has patents until 2033 at the earliest, and there are several extensions that almost everyone uses (Main 10) which are patented until even later. H.264 won't be patent free until 2027 at the earliest after some delayed patents got passed. However, even then, it depends on the H.264 revision with what extensions it has. The latest H.264 revision, Version 27, came out last year. In practice, most of the amendments and their later-expiring patents probably won't be super necessary, but some probably will be*.

 

*A big one being the addition of Scalable Video Coding (SVC) in 2007, which is used by Zoom and has patents that expire later than 2027. That was H.264 Version 8.

Link to comment
Share on other sites

Link to post
Share on other sites

18 minutes ago, Kisai said:

Honestly, there is no excuse to not have a AV1 DECODER by now

I think the biggest excuse is that Apple knows quite well what happened with the last open codec, the predecessor to AV1, VP9. It ended up only being really used by YouTube... and that's it, almost nobody else uses it for anything, even though it is theoretically competitive with HEVC for efficiency*. Sucks to add support for something and risk patent lawsuits over something only Google wants when VP9 will almost-certainly remain a fallback option for a long time yet and the M1 already has hardware decode for that. 

 

*Yes, the web has an option for a HEVC-like codec right now, and it's royalty free, but almost nobody uses it. Not unreasonable to think AV1 could end up with the same fate, ultimately.

 

In a nutshell, openness and being royalty-free does not mean adoption. In Apple's mind, VP9 was plenty proof of that. No reason to rush on the VP9 sequel. 

Link to comment
Share on other sites

Link to post
Share on other sites

14 minutes ago, gjsman said:

I think the biggest excuse is that Apple knows quite well what happened with the last open codec, the predecessor to AV1, VP9. It ended up only being really used by YouTube... and that's it, almost nobody else uses it for anything, even though it is theoretically competitive with HEVC for efficiency*. Sucks to add support for something and risk patent lawsuits over something only Google wants when VP9 will almost-certainly remain a fallback option for a long time yet and the M1 already has hardware decode for that. 

 

*Yes, the web has an option for a HEVC-like codec right now, and it's royalty free, but almost nobody uses it. Not unreasonable to think AV1 could end up with the same fate, ultimately.

 

In a nutshell, openness and being royalty-free does not mean adoption. In Apple's mind, VP9 was plenty proof of that. No reason to rush on the VP9 sequel. 

Honestly never heard of AV1, all about that H265/HEVC to take over from H264. Don’t see why apple would include an encoder that isn’t widely used. 
 

I mean even their phones use HEVC

Link to comment
Share on other sites

Link to post
Share on other sites

Disappointed that it's not a complete replacement of the 2020 MacBook Air in terms of price, I'll probably go for the M1 with double the memory for about the same price

 

I do agree that AV1 support would've been good, YouTube and Netflix (about 50% of web traffic according to some sources) use it too... better adoption than VP9

🙂

Link to comment
Share on other sites

Link to post
Share on other sites

3 hours ago, duncannah said:

Disappointed that it's not a complete replacement of the 2020 MacBook Air in terms of price, I'll probably go for the M1 with double the memory for about the same price

 

I do agree that AV1 support would've been good, YouTube and Netflix (about 50% of web traffic according to some sources) use it too... better adoption than VP9

Yeah, Netflix and Youtube use AV1 to save BANDWIDTH, not CPU power. That's generally why mobile devices need hardware support, but desktops can usually get by with software implementations.

image.png.5f0e94eca2c3de27db92d9144257367c.png

 

BTW, that is provided by Microsoft for playback. Either for free or 99 cents. If you have a RTX 30 series or an Intel 11th gen (Tiger/Rocket Lake) CPU/iGPU, and running Windows 10 1909 or later, the web browser can make use of this acceleration path.

 

So right now, Netflix or Youtube can send you AV1, but it might still be 5 years until enough CPU/GPU's have support. Apple doesn't necessarily need to be the first on the AV1 cart, but the longer they wait, the more of an issue it will be.

 

That said, watching netflix or youtube, or any video streaming service for that matter on a phone is usually painful, awful, experience. A tablet is preferable, or a smartTV, and at the price a SmartTV costs, there's really no excuse for AV1 decoding being absent.

Link to comment
Share on other sites

Link to post
Share on other sites

4 hours ago, gjsman said:

Apple actually supports VP9 hardware decode (no encode) in M1 without telling anyone.

I know. I have referenced that several times.

 

4 hours ago, gjsman said:

AV1 could be in M2 right now the same way, but I doubt that. Also, it is completely possible M2 has AV1 but without OS enablement, only time will tell.

Yeah but I doubt it as well.

 

 

4 hours ago, gjsman said:

I looked into it and you are correct on this point. I may have gotten confused and thought it was twice as slow as x265 when it was actually x264 but competitive with x265, or something like that. My bad.

It's fine. It's a simple mistake. Sorry if I came off as rude.

Anyway, some newer tests actually shows SVT-AV1 pulling ahead of x265 when it comes to encoding speed at the same quality and file size. Here are some results using SVT-AV1 version 0.8.8 (which are missing some major improvements compared to version 0.9 and 1.0). The results are from the reddit user dextorious.

 

 

4 hours ago, gjsman said:

1. Sisvel has patents from non-AOM members which may be legitimate. Because the patent owners are not in AOM, they could be used for litigation. 

Well they might be, or they might be bogus. As far as I know, Sisvel has not gone after anyway, or even disclosed which patents they have. 

In any case, regardless of how valid or not the patent situation is for AV1, it is much better than HEVC. I don't even think it is possible to get a proper license for HEVC. In order to use HEVC (legally, without any risk of being sued), you have to pay to 4 separate patent pools, as well as at least 8 different independent companies that are not affiliated with any patent pool.

You have to pay the HEVC Advance patent pool, you have to pay the MPEGLA patent pool, you have to pay the AlphaDigitech patent pool as well as the Velos Media patent pool. On top of those, you have the aforementioned companies that do not belong to any patent pool such as Qualcomm, Nokia, Intel, Lenovo, etc. And to make matters worse, the patent pools are constantly changing. Thought you were safe from being sued by Qualcomm because you paid the Velos Media license pool? Sorry but Qualcomm left that pool last year so now you have to negotiate licensing agreements with them independently.

 

The only reason why HEVC works at all for companies like Apple is because they can threaten to sue companies for other stuff if someone goes after them. The licensing situation for HEVC is honestly such a fucking joke basically anyone can sue anyone at any time. The only thing stopping a complete disaster is the fact that everyone can sue everyone.

 

4 hours ago, gjsman said:

2. Apple operates more internationally than most companies. This could end up like MPEG-2, which has 2 patents only valid in Malaysia but MPEG-LA is still extracting royalties on DVD players for them to this day. If Sisvel has even one valid patent, it needs a royalty, possibly on all devices regardless of country of sale (again like MPEG LA's continued patent fees for MPEG-2 even in the US for those Malaysian patents).

But again, Apple is already in this situation with HEVC, even if they pay license fees to multiple patent pools. AV1 is still way smaller of a risk than HEVC is.

Also, I think it is important that Sisvel has threatened lawsuits regarding VP9 since 2013 and to my knowledge there have been 0 lawsuits launched. The only one who has tried to sue because of VP# and AV1 is Nokia. They tried to sue because of some VP8 patent, and they lost...

It's been 9 years and Sisvel has still not done anything. At some point people have to call their bluff and stop giving them attention.

 

 

 

4 hours ago, Imbadatnames said:

Honestly never heard of AV1, all about that H265/HEVC to take over from H264. Don’t see why apple would include an encoder that isn’t widely used. 
 

I mean even their phones use HEVC

HEVC is dying. Only a few companies support it and even they are fairly reluctant and are trying to move away from it. The licensing killed it. 

Meanwhile, AV1 is getting a ton of support from everyone. Pretty much all big companies are either supporting it already, or plan on supporting it. 

Link to comment
Share on other sites

Link to post
Share on other sites

17 hours ago, Quackers101 said:

really great to see apple pushing a password manager. while it's unsure how good it is, unless someone has experience with this one?

 

FWIW, Anthony seemed really pleased with it in his video covering the announcements.,,I got the impression from him that it's a user friendly implementation of the thing the hard core linux geeks use.

🖥️ Motherboard: MSI A320M PRO-VH PLUS  ** Processor: AMD Ryzen 2600 3.4 GHz ** Video Card: Nvidia GeForce 1070 TI 8GB Zotac 1070ti 🖥️
🖥️ Memory: 32GB DDR4 2400  ** Power Supply: 650 Watts Power Supply Thermaltake +80 Bronze Thermaltake PSU 🖥️

🍎 2012 iMac i7 27";  2007 MBP 2.2 GHZ; Power Mac G5 Dual 2GHZ; B&W G3; Quadra 650; Mac SE 🍎

🍎 iPad Air2; iPhone SE 2020; iPhone 5s; AppleTV 4k 🍎

Link to comment
Share on other sites

Link to post
Share on other sites

I was surprised to not see much discussion here about the new Metal 3 features:


MetalFX:
   upscaling not sure yet if this is a DLSS or more AMD FidelityFX we will see (however I did see some mention it is using the NPU)
Fast resource loading:
   Think MS direct storage but even lower latency due to being on package the gpu is able to issue load/store IO commands directly to the SSD controler and even make use of the on die decompression hardware to de/compress multiple different formates... 
Mesh shaders:
  
while this was possible in Metal2 using a compute shader followed by a render pass that meant saving the result of the compute shader back to system memory then pulling it back in for vertex, tiling etc. Mesh shaders now can run at the start of the reneder pass giving the resultant modified mesh data directly to the tiler so as to avoid massively reducing bandwidth usage (really helps). Its quite interesting from a gpu design perspective that these results will be fed directly into the tiler and then fragment culling meaning many of them will never be written to vram in any way, this means you could have scenes with rather complex generated meshes without much perf impact on the rest of the render pipeline.
* Hardware RT scheduling:
 
in metal 2 there were a load of GPU based RT operations, building structures, interaction tests in bulk or from within compute or even render stage shaders was possible however to display new bulk interaction tests you needed to issue new commands from the cpu this adds quite a bit over overhead and makes building an optimal RT core (for something like blender) quite complex as you need to ensure you dispatch new work to the gpu fast enough but also want to be smart about the work you dispatch (only asking for new rays to be tested were you need rays) so you want to look at the last result before issuing new ray tests this leads to the GPU sitting ideal waiting for work to be issued, (you can see this if you attach to blender on M1 Max gpus) with Metal 3 devs can now dispatch new RT batch operations directly from within a metal shader, so you can build a compute kernel that dispatches work, looks at results then dispatches more work without needed to wait for the cpu to send new work this should reduce the wait time a lot!


 

Link to comment
Share on other sites

Link to post
Share on other sites

On 6/7/2022 at 11:38 AM, saltycaramel said:

 

What if some Gen-Z completely skips owning a traditional Pc/Mac and just goes with iPad Air M1 + usb-c hub + cheap external monitor + mouse&kb

Or, literally every med student ever.

ANECDOTE ALERT.

One of my family members studying medicine has the exact same setup, with the difference being that they have a 3rd gen iPad air and an apple pencil, usb dock and a Logitech mouse and keyboard, and they love it. Not only that, 70% of every med student that I have met has some sort of an iPad and they love it.

/ANECDOTE

 

The age of the PC being a work horse is gone, and if apple just inches closer and closer to getting the iPad up to parity with the PC.

 

Only thing is that it still feels hampered in many ways. Apple's progress with the iPad  reminds me of that Metallica lyric "so close, no matter how far".

Link to comment
Share on other sites

Link to post
Share on other sites

27 minutes ago, WolframaticAlpha said:

Only thing is that it still feels hampered in many ways. Apple's progress with the iPad  reminds me of that Metallica lyric "so close, no matter how far".

 

I like that in the years we’ve seen two ways of approaching this from opposite directions.

 

Windows 8: slap Metro and tablet mode on top of a pre-existing desktop OS.

 

iPadOS: gradually add desktop-like features on top of a pre-existing tablet OS, but never betray the spirit of the original “easy mode” for users that don’t want to use (or even discover) the “hard mode”, never making those “easy mode” users second class citizens or an afterthought.

 

Needless to say I think the latter approach makes for a better actual tablet for actual touch interaction and actual tablet software. Then you gradually add desktop-like features on top of that.

Link to comment
Share on other sites

Link to post
Share on other sites

20 hours ago, Quackers101 said:

really great to see apple pushing a password manager. while it's unsure how good it is, unless someone has experience with this one?

Macs and iOS devices already had a built in password manager for many years and it’s called iCloud Keychain. What the W3C and Fido Alliance together with Microsoft, Google, and Apple are using is making the password more of a technical element rather than something a user can see. Instead of logging in with a username and password that can be stolen through phishing, I can either use biometrics (Touch ID or Face ID), phone authenticator that generates or a security key like a Yubikey that generates a one time login. Even if the website gets hacked, what the hacker will get is a bunch of single use keys that are useless. Since it’s an open standard, it will work across devices regardless of the OS.
 

At the moment you now have the option to remove your Microsoft account password and use exclusively the Microsoft authenticator app. When you enroll to Google Advanced Protection for free, you can either use your phone or a hardware security key for verifying a login. I enabled Google Advanced Protection on my personal Gmail after being a victim of a sophisticated phishing email that promoted my bank to call me in the middle of the night. 

There is more that meets the eye
I see the soul that is inside

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now


×