AMD Intentionally Crippled Their HDMI Adapters

[microcode]

* HDMI audio is disabled on DVI outputs by default, since there is no "DVI audio" standard and thus no guarantee that all available monitors can deal with it.

Actually, DVI is perfectly capable of this, and the adapters handle it just fine when you use the FOSS drivers... I believe DVI even has a specification for doing TCP/IP over DVI(just like HDMI does).

[tstrunk]

First of all: I agree that locking out third party adapters just for the heck of it is not good, but:

[..], the cards are sold with AMD DVI->HDMI adapter, so, the user of such video cards can use the HDMI as intended...there is no anti-competitive behavior because they supply the adapter anyway.
But again, i might be missing something here...[..]

We are all missing something here in my opinion. The cards are always sold with the required adapter. What does AMD gain from locking out other adapters? Money: not really, because you get the adapter anyways. It's not like Apple, where you have to buy the cable afterwards for a lot of money. They sell one more i2c chip, but I guess that can't be it. DRM? hardly, because HDCP is there for the video anyways.

Still my money is on some DRM reason.

Edit: DHCP -> HDCP

[blackout23]

What now AMD fanboys? Even RMS recommends NVIDIA hardware, because at least the nouveau driver is really open!

http://stallman.org/to-4chan.html

[blinxwang]

What now AMD fanboys? Even RMS recommends NVIDIA hardware, because at least the nouveau driver is really open!

http://stallman.org/to-4chan.html

If you had done more than just skim the article you'd know that this malaise only affects the closed-source Catalyst drivers and that the FOSS drivers remain unaffcted. With the recent release of HDMI programming documentation by AMD, more cards will be liberated from this condition.

[blackout23]

If you had done more than just skim the article you'd know that this malaise only affects the closed-source Catalyst drivers and that the FOSS drivers remain unaffcted. With the recent release of HDMI programming documentation by AMD, more cards will be liberated from this condition.

Well but the open source driver isn't really open source. I thought that's why the fanboys always went crazy about AMD, because they're so open you know!

[bridgman]

If you had done more than just skim the article you'd know that this malaise only affects the closed-source Catalyst drivers and that the FOSS drivers remain unaffcted. With the recent release of HDMI programming documentation by AMD, more cards will be liberated from this condition.

The recent HDMI documentation wasn't related to this -- the relevant releases were maybe a year ago, and probably would have been 6-9 months before that if HDMI licensing hadn't been an issue. The adapters were an easy and effective solution -- finding a solution that worked for open source drivers was a lot harder and took a lot more time.

[rohcQaH]

What does AMD gain from locking out other adapters?

Nothing. Again, I don't think it's about detecting whether the adapter is from AMD or a competitor. It's about detecting whether the output is DVI or HDMI. And only AMD's own adapters contain the required bits for detection.

Other adapters work with the full DVI featureset. For HDMI features (i.e. audio), you need a native HDMI port or a DVI port that knows that it's attached to a HDMI device.


No, I don't know why AMD wants to avoid sending HDMI audio over DVI ports. Marketing, spec adherence, support considerations, licensing or other contracts are just as likely to be the reason. It's not an attempt to make money by monopolizing the adapter market.

[Zajec]

Oh, now I'm scared with the flame caused by my discovery/e-mail...

...wait, you guys didn't know about this? This is really, really old news we're talking about here...

For me the most interesting part is *how* ATI detects manufacturer of the adapter. Before finding that post on avforums I couldn't imaging how it's possible to "talk" with the adapter. I expected it to just pass the DVI pins to HDMI pins, nothing more. Finding out it's a matter of I2C was pretty interesting for me
Btw, try to Google for a moment about ATI HDMI to DVI adapters. You will find bunch of nonsense answers that they have extra pins for audio handling

Wait, I don't get it... So if you have a monitor that only has HDMI, and you have a card that doesn't have any HDMI slots, and you have the ATI DVI in-HDMI out adapter, then you can get audio over the DVI slot? Just how often does such a situation occur?.. And how many "cheap" DVI in-HDMI out adapters even support audio over DVI, given that it's not exactly standard and that not a whole lot of graphics cards support such a thing to begin with? In fact, if a card has no HDMI slots, why would it have an integrated sound card to begin with?..

First of all, DVI and HDMI transmit the same (TMDS) digital signal. The difference is that specification for DVI doesn't include sending audio bits between the frames. So ATI decided to implement feature of sending HDMI-standard signal over DVI port. Pretty clever.
DVI to HDMI adapters don't have to support audio. They just pass the pins from one output to another.
Yes, many ATI cards don't have HDMI output, but have audio engine. For example my HD4850 has 2xDVI (no HDMI) and can send audio just fine.

So, did I get this right?
* HDMI audio works fine if you just use a HDMI cable in the HDMI port. As expected.

Right.

* HDMI audio is disabled on DVI outputs by default, since there is no "DVI audio" standard and thus no guarantee that all available monitors can deal with it.

Not really. No driver blindly sends a HDMI signal. First it checks TV's EDID to see if it (TV) supports audio.

* If you wish to use your DVI output to attach HDMI devices, AMD has made some special converters that, if detected, treat the DVI output as a HDMI output and will enable HDMI-only features such as audio.

Yes, but as I said, it's not done unconditionally. I'm sure fglrx first check if TV can understand HDMI signal (with extra frames).

It's not a common problem, either. If you need HDMI audio, you can just use the HDMI output, which virtually all newer cards have. Or you just use the adapter that was included with your box at no extra charge.

That's right. My only problem is my friend lost his original ATI adapter for his HD4850.

Unless there is something i didn't understood correctly, the cards are sold with AMD DVI->HDMI adapter, so, the user of such video cards can use the HDMI as intended...there is no anti-competitive behavior because they supply the adapter anyway.

But again, i might be missing something here...

Yeah, I don't believe it was done for $. Especially (as you noticed) as adapters were usually/often included.

[Zajec]

So as you can see, it's unlikely that ATI wanted to earn some extra money or really meant to be evil for the users. I think there must some some other explanation. Overall, ATI adapters were almost always included, so it was really nice from them.

I think it may be somehow related to the HDMI licensing issues. That sounds pretty likely for me, but to be sure, you have to ask AMD.

[Brane215]

Article is quite shallow or borderline stupid.

Who would use elcheapo, trivially copiable serial eeprom to lock you into anything ?

It is obvious it is there for some other reasons. Like to differentiate interfaces with decent enough signal integrity or to signal to PC to send the audio ( someone mentioned EDID) etc etc.