I hope in the future that Yamaha and Sound United engineers tell their respective marketing departments to go take a long walk off a short pier. History has shown time again that rushing products to production always causes problems and damages the companies' reputations. I would also like to see the marketing management to be relieved or demoted from their positions for short sightedness and incompetence.
I agree with you, but it is more complex. Please, read on. I would not scapegoat marketing teams for short-sightedness and incompetence. They did as they were told to do. The issue was deeper, across departments, and in management. Lessons definitively need to be learnt from this. It did not look good at all. And it did not need to be that way, had they paid attention to details of HDMI spec while testing with real devices. It's interesting to understand how this process unraveled, and why testing & certification are so important. I will give one example from Yamaha and SU.
The knowledge below was put together by tech community after several months of analyzing what was going on. The companies have never confirmed to the public real reasons, but one can try to put a rough jigsaw puzzle together with tech enthusiasts and industry insiders from various backgrounds.
The two companies had two major problems with HDMI chips. There were only two chips on offer back then:
1. for Sound United - one fast input port with 40 Gbps
2. for Yamaha - multi 24 Gbps ports (over-reliance on DSC to compress higher bandwidth video image)
A product manager needs to be able to have an overall grip on product development. The product was developed before Covid struck and showed behind the closed doors at CES 2020. Product leader is in charge of collaboration, accountability and all teams working smoothly together, from marketing to engineers, to software developers, to procurement, to QC. Leader also need to have a minimal level of tech knowledge of current and near-future consumer electronics market, major popular devices and imagination how features work together when devices are connected. Or, at least, to be briefed about it in reasonable detail by relevant teams. Engineers need to know the HDMI spec inside out. Procurement team needs to know what chips to order, how to communicate feature requirements during sourcing, why those chips and not others.
If several teams do not speak the same language and do not exactly understand how product and features are supposed to work in a repeater mode with
new tech (because they did not test it in real life), you get faulty chips on faulty boards in retail. It was a collective responsibility for the failure with a series of bad decisions. Somebody also decided
not to test those AVR machines with pre production samples of GPUs and consoles. Those were available. Had they asked Microsoft, AMD, Nvidia or Sony to send them a sample unit for testing, they would have quickly discovered in their AV lab that AVRs did not work properly. It was not difficult to do in-house connectivity QC - send 4 emails or make 4 phone calls and get samples to test.
Let's go into some detail.
FRL data protocol is in the heart of HDMI 2.1
AVRs from Sound United failed to establish FRL (fixed-rate link) data rate with XboxX and TV (4K/120 signal). XboxX worked properly directly with TV. The same was the case with Yamaha's chips, but they never rolled out any firmware to owners to test this, as everybody knew what was happened with SU's devices. This is the main reason why Sound United was more in the spotlight at that time. Yamaha's AVRs were selling in 2020 too and those worked with older HDMI 2.0b TMDS signalling, up to 18 Gbps. Officially, their new AVRs did not fail, as they announced future firmware to enable FRL signals and other features. It is a matter of debate in tech community whether they knew at that moment that their original boards were not going to deliver 24 Gbps link. Anyway, a few months later, they ordered new boards with working 24 Gbps chips for 2020 models, the chips that could do FRL over 24 Gbps. They also ordered new boards for 2021 models; faster 40 Gbps boards. The public did not know this differentiation on slower and faster boards until August 2021. Here is the diagram how HDMI 2.1 signal works over 4 data lanes. Fourth lane (TMDS clock) is repurposed for another FRL data lane, enabling higher bandwidth, up to 12 Gbps per lane. 2020 models: 4x6 Gbps, 2021 models 4x10 Gbps (Nvidia GPU 4x12Gbps).
A simple example of group incompetence, or lack of imagination, or naive ignorance, or blindness, if you will, was initial choice of 24 Gbps chips for entire Aventage line-up that arrived at CES in January 2020. In official HDMI 2.1 spec, there is a new feature called DSC (display-stream compression). This feature has been used routinely for several years over DisplayPort on monitors. It finally came to HDMI 2.1 too. If a device supports DSC protocol, then 24 Gbps chips should bring enough bandwidth to compress any signals that need more than 24 Gbps. Compression ratio is up to 3.75:1. So, in theory, DSC on 24 Gbps chips could compress up to 80 Gbps video signal into 24 Gbps pipeline, which is an image up to 8K/60 10-bit RGB/444 (see in the last chart below).
Plenty of bandwidth to work with, one would say. This would give new AVRs a wide range of compatibility and future-proofing. Effectively, you can connect any wide bandwidth source and display to it and it should work. Nothing to worry about, as compression over 80 Gbps signal will never be needed in next 10 years. So far so good. Now, for DSC to work,
all devices in the chain must support it. See below.
You either have DSC signal end-to-end or uncompressed signal end-to-end. DSC and non-DSC cannot operate at the same time. DSC encoder is at source device and DSC decoder is at the sink. AVR needs to unpack audio from it and play, and pass whatever video is there, but the signal must be
consistent across the pipeline. What nobody from Yamaha checked, or decided to ignore, was whether source and sink devices in consumer market widely support DSC feature. Critics might argue it was not their job to do that. Well, you intend to sell a device that is supposed to work with dozens of models of TVs, monitors, projectors, diverse sources, etc. that hundreds of thousands of people have at homes. It might be wise and useful to know about DSC compatibility and research DSC support on TVs' SoCs.
It turns out that few HDMI 2.1 devices support DSC, even today. Apart from a handful of high-end TVs, an odd new monitor, Nvidia GPU and consoles, majority of current 4K TV that we have at home do not have DSC decoder in their SoC. One of reasons for this is to reduce additional data processing over DSC. TV's SoC has already a lot of jobs to do to display images. Another layer of data stream makes processing more complex. Connect and you get black screen or fall down to image that can pass through lowest bandwidth available on device in the chain. I am yet to see any meaningful list of TVs and monitors that support DSC over HDMI 2.1. It does not exist.
Below is DSC chart to show different compression ratios for different video signals, courtesy of a tech specialist form another forum.
Why would anyone release three AVRs with 24 Gbps chips if you know that majority of TVs on the market do not support DSC feature? It is AVR that needs the widest bandwidth in order to be widely compatible. You can release a mix of 40 Gbps and 24 Gbps ports, as Onkyo did, to save some costs on chips. It makes sense.
Another thing is image quality for uncompressed data. Critics may argue that a difference in image quality between signal that could fit into 24 Gbps and 40 Gbps is not that noticeable. This depends on content used and it's mostly related to text:
24 Gbps - 4K/120 10-bit 4-2-0 - major impact on PC use: more blurry text and less sharp letters in live content in games, files and browser, more so on OLED screens.
40 Gbps - 4K/120 10-bit RGB/444 - scrolling and text are sharp and smooth, colors better in places
That's the story for tonight. The more details we know and share, the better we understand wider picture.
