AMD Intentionally Added Artificial Limitations To Their HDMI Adapters

[Sionyn]

AMD has quietly been adding an extra EEPROM chip to their DVI-to-HDMI adapters that are bundled with Radeon HD graphics cards. Only when these identified adapters are detected via checks in their Windows and Linux Catalyst driver is HDMI audio enabled. If using a third-party DVI-to-HDMI adapter, HDMI audio support is disabled by the Catalyst driver.

http://lists.freedesktop.org/archives/dri-devel/2013-October/046676.html

[mariush]

I don't see the problem. If the card has no hdmi connector on the bracket, you get a DVI to HDMI adapter with the video card.

[AndrejaKo]

I don't see the problem. If the card has no hdmi connector on the bracket, you get a DVI to HDMI adapter with the video card.

But that doesn't always happen with all cards and sometimes a computer may come with a card installed and no accessories provided. Then the problem suddenly appears.

[wraper]

I think that's because DVI does not support audio. They might be doing this to avoid compatibility problems. Because with generic adapter video card does not know if there is HDMI device connected, not DVI. But DVI monitor could work incorrectly if there is audio stream transmitted.

[ve7xen]

They still make cards with DVI?

[AndrejaKo]

As far as I know, DVI is the current video standard for connecting monitors to computers. Are you sure you haven't confused it with VGA? The other competing standard seems to be DisplayPort, but from what I can see, it's still not as popular.

[ve7xen]

As far as I know, DVI is the current video standard for connecting monitors to computers. Are you sure you haven't confused it with VGA?

I haven't purchased one in a while, but last time I did I noticed a proliferation of HDMI and DisplayPort on most cards. Figured they would've killed DVI off entirely by now.

[TMM]

HDMI and Displayport have almost completely replaced DVI in consumer grade monitors and graphics cards.

DVI is still popular for business/industrial computers where life cycles tend to be longer and new computers are purchased while retaining old monitors, or new monitors are purchased for existing computers.

[amyk]

HDMI and Displayport replacing DVI? Another case of "big durable connector gets obsoleted by tiny flimsy one"...

Am I the only one disappointed that neither OP's link nor the forum linked to therein contained pictures of the insides, EEPROM data, or any other interesting bits...?

[peter.mitchell]

It's probably because of the Digital Restrictions Management (DRM) in HDMI. They don't want you to be able to output uncompressed full quality audio if the BluRay disc says you are not allowed to. It's part of the certification requirement for the Secure Audio Path shit built into the standard.

Basically they want to verify that their own converter is in use and no one modified to capture the audio stream somehow. Bullshit of the highest order.

Most likely, as a part of the hdcp thingo

[andtfoot]

HDCP support can be enabled on DVI interfaces so I'm not sure why that would be the case.

I would have thought a lack of audio support of the sink (e.g. plugging into a DVI display/sink) would be picked up by EDID, so I don't know about that theory either.

Both are definitely still possible; I'm not ruling them out at all.

The only thing I can think of otherwise is some sort of loophole in the licensing for the HDMI standard. If the port doesn't act like a HDMI port normally, maybe they can avoid the fees. 

[rr100]

HDMI and Displayport replacing DVI? Another case of "big durable connector gets obsoleted by tiny flimsy one"...

Just what I was thinking, I have a video card that has both DVI and HDMI and I have to use both, with 2 monitors. Video cables aren't of course thin and flexible and each time I have to fiddle with some other cables at the back of the computer the flimsy HDMI loses connection, desktop gets rearranged on one monitor and of course I go @@#@#@!@@@@!!

[andtfoot]

HDMI and Displayport replacing DVI? Another case of "big durable connector gets obsoleted by tiny flimsy one"...

Just what I was thinking, I have a video card that has both DVI and HDMI and I have to use both, with 2 monitors. Video cables aren't of course thin and flexible and each time I have to fiddle with some other cables at the back of the computer the flimsy HDMI loses connection, desktop gets rearranged on one monitor and of course I go @@#@#@!@@@@!!

As someone in the commercial Audio Visual industry, this is one of the aspects of the often bemoaned HDMI that causes some grief.

There have been a few attempts by manufacturers to fix this.
Some have retention screws on their equipment, and then provide adapters to attach to the connector like this:

http://www.extron.com/product/product.aspx?id=dtphdmi230rx&subtype=360&s=3
See the screw above the connector. It attaches the below to the equipment.
http://www.extron.com/product/product.aspx?id=lockit&subtype=362&s=3
The cable is tied to this.


I've also seen (but not tried) an attempt to put a locking mechanism in the connector itself:

http://www.madisontech.com.au/broadcast-av/brand/perfect-path
(the manufacturer website is down at the moment)

[sub]

This was discussed on Slashdot a few days ago; there it was claimed that the DVI standard does not have any provision for audio signals.  The idea, it seems, is that they check to see whether it genuinely is a DVI->HDMI adaptor before sending non-standard signals out willy-nilly.

[daveshah]

If anybody is curious, I just disassembled an old AMD adapter I had lying about - photo attached.
The device used is a 24C164WI from CSI. The closest I could find is this part from mouser: CAT24C164.

For some reason, this part is a 2Kbyte part, which seems odd considering its purpose. This made me consider some kind of bizarre security/encryption system, but upon dumping it appears to be mostly empty - it could definitely fit in a 24C02. Considering the production volumes, I would have expected the bean counters to make this substitution. I only presume they had a large surplus of this part from another product, or found a very cheap source.

Also attached is a dump, in case anyone wants to make a clone (also note the chip lives at 0xA0-0xAF, implying A0, /A1 and A2 are low). The interesting part is only 17 bytes long, and consists of:
0x41 0x4D 0x44 0x01 0x00 0x0B 0x36 0x31 0x34 0x30 0x30 0x36 0x33 0x35 0x30 0x30 0x47

This is the text 'AMD', then 3 separator bytes, then the text '6140063500G'. This almost matches the serial number embossed on the housing, which is '6140063501G'. Just to be sure this wasn't a read error, I checked again, so perhaps the machinery at the factory got out of sync?

I am curious as to whether this number is used for anything, and whether it is unique per device. If anyone else can tell me what number is embossed onto their adapter, I would be very interested.

[Marco]

They have no commercial advantage to block "competitor" adapters, they have no competitors, because there is no adapter market as far as they are concerned.

This isn't like say Apple AV adapters where Apple treats them as money makers.

[andtfoot]

Also attached is a dump, in case anyone wants to make a clone (also note the chip lives at 0xA0-0xAF, implying A0, /A1 and A2 are low). The interesting part is only 17 bytes long, and consists of:
0x41 0x4D 0x44 0x01 0x00 0x0B 0x36 0x31 0x34 0x30 0x30 0x36 0x33 0x35 0x30 0x30 0x47

That is curious. That's the same address that the standard EDID chip lives at.

Given the offset of the data (518 bytes), I can only presume that they are giving a wide berth around the standard EDID data block(s). It would allow the 128 byte VESA block, plus 3 extra 128 byte extension blocks. I've only ever seen one extension block being used though.

I'm guessing that the conflict between the two slaves when reading the EDID would be handled by I2C arbitration with the adapter's chip being left floating high (filled with 0xFF where it's blank), and the monitor's chip taking priority because it is sinking the line to ground. This scenario would be reversed when the special AMD data is being read.

I'll have to dig and see if I can find an adapter floating around.

[daveshah]

That explains why they used a larger EEPROM. I suspect the reason for doing this is to make accessing it in software easier.

[mariush]

Yeah, I can confirm the serial number... the adapter below is from my old sapphire hd4850, still in its original packaging

[peter.mitchell]

Yeah, I can confirm the serial number... the adapter below is from my old sapphire hd4850, still in its original packaging

I have boxes of these friggin things, so i was surprised someone made a fuss about it, I have 100+ dvi-> vga and dvi-> hdmi adapters, just as many of the YPrPb breakout cables...

[amyk]

Thanks for the teardown

That's not a serial number, it's a part number as you'll find out if you Google it - the '0G seems more common than the '1G, I suspect the latter is just a physical revision and the EEPROM contents remain the same.

If the device's EDID EEPROM is smaller, wouldn't it wrap around and act as an AND-mask to the data in the adapter?

[dr.diesel]

HDMI and Displayport have almost completely replaced DVI in consumer grade monitors and graphics cards.

I don't believe any graphics card manufactures today are using the HDMI v1.3 spec, which means your limited to 1900x1200 max resolution. 

I was quite thankful for DVI on my Nvidia 660 when I got my 2550x1440 monitor!

[peter.mitchell]

HDMI and Displayport have almost completely replaced DVI in consumer grade monitors and graphics cards.

I don't believe any graphics card manufactures today are using the HDMI v1.3 spec, which means your limited to 1900x1200 max resolution. 

I was quite thankful for DVI on my Nvidia 660 when I got my 2560x1440 monitor!

2560x1440 is where it's at! Some people go for multi monitor,others go for high res 27 inch screens, i go for both;
https://dl.dropboxusercontent.com/u/677635/IMG_20130724_002502.jpg

[NiHaoMike]

In practice, I have seen few great implementations of HDMI audio. The speakers integrated into most TVs are rivaled by cheap computer speakers and passing the HDMI through a receiver requires the receiver to be on (and using more power than many smaller displays and low power PCs!) just to get an image, plus there's another possible source of lag. Most TVs do have S/PDIF outputs, but then you have to have the TV on to get audio, which you probably don't want when you just want some background music.

[andtfoot]

In practice, I have seen few great implementations of HDMI audio. The speakers integrated into most TVs are rivaled by cheap computer speakers and passing the HDMI through a receiver requires the receiver to be on (and using more power than many smaller displays and low power PCs!) just to get an image, plus there's another possible source of lag. Most TVs do have S/PDIF outputs, but then you have to have the TV on to get audio, which you probably don't want when you just want some background music.

There is mostly likely cheaper versions of this, but here is a device that may solve that issue:
http://www.extron.com/product/product.aspx?id=hae100&s=0