News China launches HDMI and DisplayPort alternative — GPMI boasts up to 192 Gbps bandwidth, 480W power delivery

Status
Not open for further replies.
If the cable has network capabilities or attempts to connect to the internet for telemetry it's not impossible. https://www.pcworld.com/article/241...sly-watch-your-screen-via-hdmi-radiation.html
What cables are connecting to the internet for telemetry? Displayport to my knowledge don't support the kind of connectivity you'd need for that. HDMI does support ethernet and you could attach a small, malicious ethernet device... but no device I've seen has had the option to bridge network connections to HDMI, much less defaulted to it (and a cable that sticks a malicious device here might break ARC). Thunderbolt or USB4 cables could secretly include a hub and hang a malicious device off the hub... but it's gonna show as a hub and have to load drivers on the host system to enable the device to talk to the internet, and being asked to load an unsigned driver for your USB cable should be the biggest of red flags.

I have no idea how GPMI would implement a backdoor that wouldn't be immediately obvious to Intel/Nvidia/AMD/Qualcomm (or people making devices with their chips).

While theoretically possible, that is not a realistic threat at this time for anyone not being targeted by state-level actors.
 
What cables are connecting to the internet for telemetry? Displayport to my knowledge don't support the kind of connectivity you'd need for that. HDMI does support ethernet and you could attach a small, malicious ethernet device... but no device I've seen has had the option to bridge network connections to HDMI, much less defaulted to it (and a cable that sticks a malicious device here might break ARC). Thunderbolt or USB4 cables could secretly include a hub and hang a malicious device off the hub... but it's gonna show as a hub and have to load drivers on the host system to enable the device to talk to the internet, and being asked to load an unsigned driver for your USB cable should be the biggest of red flags.

I have no idea how GPMI would implement a backdoor that wouldn't be immediately obvious to Intel/Nvidia/AMD/Qualcomm (or people making devices with their chips).


While theoretically possible, that is not a realistic threat at this time for anyone not being targeted by state-level actors.
Is a load of paranoia, such attacks require a relatively close range to be feasible. And wouldn't require entirely new technology or standards like this cable to do. But I guess people want an excuse to spout anti-Chinese nonsense.
 
Has anyone seen this in action? If lno one adopted, produced or mentioned implementation specs other than x GB/s and 480w can we even consider this real?

I am releasing a standard that'll carry 4x times what this cable can in data and power. It will go twice as far too. I also promise 2kw on 30 gauge wire.

It's going to be so OP. Seriously though, 480 w sounds great until you think about how thick the cable needs to be at any distance and how rare the power electronics that handle that are. They could try to double the max USB PD voltage I've seen, but the power fets that handle that get expensive. Huge cable or high price? Or max length limit of like a foot and still decently thick wires? All sound gross.

And either one to what? The 1 other device that this ends up implemented on?

I could believe if they adopted usb PD and the 20v that tops at. I seriously doubt anyone's going to be happy with how thick a 480w version of that ends up being

My best guess is we never see this so called standard in the wild. If we do it won't be as stated here. I most seriously doubt the power claim.
 
Last edited:
Given the trouble we have with signal integrity for 80Gbps I'm sure neither of these will function without having power requirements. Pretty much every USB-B implementation had nightmare connectors, but I suppose given the size it would be possible to have more robust internals. It's possible they could have gone with a hybrid optical/copper design for data/power as well (none of the articles I've seen mention specifications).

Needless to say while it's not impossible the claimed data rates don't seem particularly believable especially if it's full duplex.
 
Last edited:
  • Like
Reactions: gg83
Oh, I hope they don't mess with USB-C . We don't need another non-standard USB-C like Oppo's phone charging cables.

Also, what is the range of (passive) GPMI cables? (Imagine the use-case of hooking up a ceiling-mounted projector.)

HDMI-CEC is quite limited, and every brand seems to have its own extensions, making only a small subset available between devices from different brands. Professional monitors and projectors have additional control interfaces over RS-232 or Ethernet, and all of those are of course proprietary protocols too.
I would like to see the control signals in GPMI being a proper standard.

... and of course: By "8k" do they actually mean 8 K or something closer to 7.5 K ?
 
One of the nice things about how Chinese tech has so far developed is their adoption of the existing computing standards, thus facilitating interoperability. If China goes its own way, with many of the standards used in computing, the West is going to eventually lose. China already undercuts Western computing products in developing countries and will quickly dominate the markets in most countries, outside of North America and Europe.

If Western manufacturers want to sell into those markets, they'll have to adopt the same standards. At that point, it's only a matter of time before those same products start being sold domestically. Then, we will have gained nothing by trying to shut out China from involvement in our computing standards and will have lost the opportunity to participate in shaping their continued evolution. Let's hope we don't also face discrimination in licensing fees, although I have no reason to expect we wouldn't.

I've been hoping this wouldn't happen, but it seems now to be inevitable. Next up: a PCIe/CXL replacement? Or how about USB, like @Findecanor mentioned?
 
Looking at some of the math: For a 5' cable at 12V and 480W you're going to want an 11 AWG wire, or ~2.3mm. Those pins look a fair bit smaller than that, even if the current is split between a few of them.


https://www.inchcalculator.com/wire..._drop=5&uc_gauge=0-awg&uc_calculator_type=mws

https://www.inchcalculator.com/wire...e=sp_ac&uc_gauge=11-awg&uc_calculator_type=wd
USB-C is 48V for the 240W power delivery so I'd assume these would also both be 48V rather than 12V.
 
How does the power delivery going to work? The cable will be connected to a display port whether it's PC or console. RTX5090 is sucking 600W and melting connectors so we are adding another 480W? (my LG OLED TV max power draw is north of 300W).
 
USB-C is 48V for the 240W power delivery so I'd assume these would also both be 48V rather than 12V.
I'm not the only one thinking that, https://www.reddit.com/r/hardware/c...=web3xcss&utm_term=1&utm_content=share_button

It's more type-b than type-c.

Translation from announcement webpage:
The following is a comparison of the specifications of GPMI quality HDMI 2.1 and DisplayPort 2.1:
◾HDMI 2.1 TMDS (18Gbps / No power supply)
◾HDMI 2.1 FRL (48Gbps / No power supply)
◾DisplayPort 2.1 UHBR20 (80Gbps / 240W powered)
◾GPMI Type-C (96Gbps / 240W powered)
◾GPMI Type-B (192Gbps / 480W powered)
 
We should plug these into our Nvidia graphics cards. Maybe they wouldn't melt.
This is actually the first thing i thought: with only two of these you could power a 5090 and it couldn’t possibly be worse than the standard NVidia is using now.

Maybe gpu’s are at a point where external psu’s just for the gpu make sense, then we could go back to far more modest psu’s for the rest of the system.
 
What does the type-B connector look like?
My market research indicated people love reversible connectors but need more frustration and having to look at them closely.

To this end I decided on a reversible connector with electro mechanical micro actuators that change how both the plug and sockets are keyed every 1 second.

There should be a synchronicity that allows insertion at least every 15 seconds as long as you don't have the theoretically reversible connection upside down relative to the changing key pin patterns. 👹

Also the exposed metallic strain relief is 80v above ground.
What does the type-B connector look like?
 
Status
Not open for further replies.