Thursday, June 19th 2025

AMD's Next-Gen UDNA-Based Radeon GPUs to Use 80 Gbit/s HDMI 2.2

Just as the RDNA 4 rollout cycle is completing, the rumor mill for the next generation of Radeon GPUs is intensifying. According to @Kepler_L2 on X, a reliable leaker, AMD is equipping its next-generation UDNA-based GPUs with the HDMI 2.2 display protocol, capable of supporting 64 and 80 Gbit/s, meaning that the highest-end bandwidth of 96 Gbit/s (Ultra96) is not expected to arrive until the next generation. The HDMI Forum, the consortium behind the HDMI standard, officially released the HDMI 2.2 specifications in January during CES. The HDMI forum also foresees HDMI 2.2 cables hitting the shelves sometime in Q3 or Q4 of 2025, which means the rollout is near. However, AMD's next-generation GPUs will not utilize its entire feature spectrum in the coming GFX13 GPU architecture.

The HDMI 2.2 update introduces a future-proof Fixed Rate Link architecture that delivers up to 96 Gbps of bandwidth, enabling uncompressed 4K at 240 Hz or 8K at 60 Hz with full 4:4:4 chroma, and paving the way for even higher resolutions, such as 10K and 12K at 120 Hz. It accommodates both uncompressed and visually lossless compressed formats, as well as advanced chroma sampling options. This massive data throughput meets the growing demands of immersive audiovisual experiences from next-generation gaming and augmented and virtual reality to professional light-field displays, large-format digital signage, medical imaging, and machine-vision systems by doubling payload capacity every two to three years. The Ultra96 HDMI cable is designed to handle all HDMI 2.2 features. This cable is accompanied by the new Latency Indication Protocol, which ensures precise lip-sync and audio-video alignment across complex multi-hop setups involving receivers, soundbars, and other downstream devices.
Sources: @Kepler_L2 on X, HDMI Forum
Add your own comment

53 Comments on AMD's Next-Gen UDNA-Based Radeon GPUs to Use 80 Gbit/s HDMI 2.2

#1
Chrisy
Very nice but still no working HDMI 2.1 on Linux :mad:
I know that HDMI is blocking foss but still...
Posted on Reply
#2
Unregistered
ChrisyVery nice but still no working HDMI 2.1 on Linux :mad:
I know that HDMI is blocking foss but still...
Maybe the solution could be with the open standard like GPMI Type-B?
#3
R-T-B
ChrisyVery nice but still no working HDMI 2.1 on Linux :mad:
I know that HDMI is blocking foss but still...
No working on the OSS side, which is quite bad yes, but ironically nvidias closed source driver works. I guess that's one of the few benefits of being closed.
Posted on Reply
#4
Tek-Check
BabaCan this do anything that DP2.1 can't? It's not even doing the full bandwidth similar to what AMD pulled with DP2.1 initially. So the wait begins for a slower connector than what's already out there.
HDMI can do eARC function and CEC. Small things.
DP is more versatile, of course, and can do MST.
80 Gbps on the first gen silicon is fine.

AMD pulled DP54 on client GPUs and DP80 on PRO cards in 2022 already , when NOBODY else wanted to roll out DP 2.1. Just saying, for the sake of historical accuracy.
And AMD does not charge more money for those ports. Look at monitors. Asus UCDMR with DP80 is almost $500 more expensive than its clone from last year with DP 1.4a. Insane.
Any complaints for late roll out of DP 2.1 to Nvidia and monitor vendors, please.
Posted on Reply
#5
ymdhis
BabaCan this do anything that DP2.1 can't? It's not even doing the full bandwidth similar to what AMD pulled with DP2.1 initially. So the wait begins for a slower connector than what's already out there.
Yeah, connect to AV equipment.
Posted on Reply
#6
hsew
ymdhisYeah, connect to AV equipment.
Pixelhue would disagree.
Posted on Reply
#7
TheinsanegamerN
PizderkoMaybe the solution could be with the open standard like GPMI Type-B?
Imma blow your mind here.

There is already an open standard. Displayport. It's existed since 2006. And it's available on another open standard, USB C.
ymdhisYeah, connect to AV equipment.
A/V equipment often still has VGA on it. Not our fault they're so far behind.

and displayport to HDMI adaption is trivial.
R-T-BNo working on the OSS side, which is quite bad yes, but ironically nvidias closed source driver works. I guess that's one of the few benefits of being closed.
Yes, one of the benefits of closed source is getting approval from the media cartels to stream their DRM'd content.

If only everyone just refused to use such content...
Posted on Reply
#8
R-T-B
Tek-CheckAny complaints for late roll out of DP 2.1 to Nvidia and monitor vendors, please.
Literally no one is blaming NVIDIA. We are all blaming the HDMI forum.
Posted on Reply
#9
dgianstefani
TPU Proofreader
PizderkoMaybe the solution could be with the open standard like GPMI Type-B?
What's the point? Is on virtually zero monitors/TVs, plus Displayport exists and so does Type C/Thunderbolt.

I just want AIB cards to start coming with a Type C output again, like they used to couple gens ago. Something like 2x DP, 1x HDMI, 2x Type C (DP).
Posted on Reply
#10
Event Horizon
Uncompressed 4K at 240 Hz is more than enough for me. As long as they don't break DSC like a certain competitor then I'd also be okay with DSC compressed 4k 240Hz.
Posted on Reply
#11
Centurionio
Cool but where are the 8k 144Hz monitor ?
Posted on Reply
#12
AsRock
TPU addict
ymdhisYeah, connect to AV equipment.
Yeah and seeing DP on that is going to be real exspenive or not at all. Stuck with crappy HDMI and it's HDCP BS.
Posted on Reply
#13
Dr. Dro
Tek-CheckAMD pulled DP54 on client GPUs and DP80 on PRO cards in 2022 already , when NOBODY else wanted to roll out DP 2.1. Just saying, for the sake of historical accuracy.
And AMD does not charge more money for those ports. Look at monitors. Asus UCDMR with DP80 is almost $500 more expensive than its clone from last year with DP 1.4a. Insane.
Any complaints for late roll out of DP 2.1 to Nvidia and monitor vendors, please.
It's not that they didn't want, it's that Ada physically can't do that. It's a DP 1.4a design, which means it tops out at HBR3 bandwidth.

However, the intentional market segmentation you've mentioned from RDNA 3 (which is utter stupidity in my book) carried over to RDNA 4, and the 9070 XT has UHBR20 support disabled as well. Full bandwidth DisplayPort is gated behind the Pro R9700. No Radeon Pro, no state of the art monitors for you. You literally can't make this up - there is a day that AMD won't snatch defeat from the jaws of victory, but today is not that day. And neither is tomorrow, apparently. Thankfully, there is a lot of time for AMD to reconsider this decision, if the rumor in the OP holds any truth.

There's no way this is Nvidia's fault, not in the slightest - and with their Arc A770, Intel went as far as devising a workaround for HDMI's restrictions on open source by using a hardware signal converter (the PCON). The only company that deliberately imposed any artificial limitations here is AMD, and these pertain only to DisplayPort anyway.
Posted on Reply
#14
Tek-Check
R-T-BLiterally no one is blaming NVIDIA. We are all blaming the HDMI forum.
What does HDMI Forum have to do with Nvidia's delay of DisplayPort 2.1? Nonsense. DisplayPort is managed by VESA.
Posted on Reply
#15
R-T-B
Tek-CheckWhat does HDMI Forum have to do with Nvidia's delay of DisplayPort 2.1? Nonsense. DisplayPort is managed by VESA.
Ok man, I was talking about HDMI. We seem to be having two separate discussions and they are getting crossed. No worries it happens.
Posted on Reply
#16
Bruno_O
TheinsanegamerNA/V equipment often still has VGA on it. Not our fault they're so far behind.
and displayport to HDMI adaption is trivial.
please do show me 1 single (2024+) AVR that has a VGA port.... everything is 100% HDMI based for years now

you all just love to whine and complain about HDMI vs DP, just let it go

HDMI is the standard - consoles, TVs, projectors, AVRs, etc. and it will continue to be the standard for the next decade unless something catastrophic happens. It is perfect for consumers, it's simple and does everything it needs to do.

DP doesn't even support eARC, it will never get mainstream adoption outside PC monitors... it doesn't matter that DP is superior as for consumer HDMI 2.0+ is plenty. HDMI has won this battle a decade ago

again, let it go
Posted on Reply
#17
Tek-Check
Dr. DroIt's not that they didn't want, it's that Ada physically can't do that. It's a DP 1.4a design, which means it tops out at HBR3 bandwidth.
"It's not that hey did not want... " Where do you take such nonsense from?
If they had really wanted to introduce DP 2.1 in 2022, they had all tools to do it, like AMD and Intel. Nobody was hiding new ports from them. It was enough to invest into new PCB and install the standard. Instead, they were stingy and decided to recycle the PCB from 3000 cards, thus slowing down the adoption of DP 2.1 in the entire monitor industry for two years, apart from 2-3 models that came out in 2023, such as G95NC.
In this way, Nvidia showed a middle finger to all buyers of 4000 cards. Do not defend them for not using available technology.
Dr. DroHowever, the intentional market segmentation you've mentioned from RDNA 3 (which is utter stupidity in my book) carried over to RDNA 4, and the 9070 XT has UHBR20 support disabled as well. Full bandwidth DisplayPort is gated behind the Pro R9700. No Radeon Pro, no state of the art monitors for you. You literally can't make this up - there is a day that AMD won't snatch defeat from the jaws of victory, but today is not that day. And neither is tomorrow, apparently. Thankfully, there is a lot of time for AMD to reconsider this decision, if the rumor in the OP holds any truth.
I agree that DP54 was unfortunate on consumer cards. However, there is more here to unpack. AMD was limited by the industry at that time. The best is to watch the video I linked below by Level1Tech. Wendell explains it all nicely. At the point of design of consumer 7000 cards, traditional physical DP ports offered on the market at that time did not have enough of signal integrity to sustain clean 80 Gbps connection without risking black screens. This has nothing to do with AMD. Even W7000 PRO cards with that port support DP54 only. The physical connector available at that time did not produce clean enough DP80 signal. The so called "eye diagram" was a bit foggy. Wendell shows it in testing. The only port that was able to do DP80 at that time was mini-DP. PRO cards have one such port for DP80 signal. As you know, almost no one in gaming community wants to see mini-DP port on consumer GPU, as vast majority of monitors did not have one. So, mainstream cards offered full size DP ports with DP54. It was not segmentation. I hope this explains it.

I do agree that 9000 series could have upgraded to DP80, but I also understand why they did not do it. They offered DP 2.1 ports two years before anyone else in the industry, but monitor vendors decided to largely ignore it, wait for Nvidia 5000 cards and save costs on display IC. Even today, DP 2.1 monitors are rare, and of those on offer, several feature subpar DP40/DP54 where DP80 bandwidth is needed for uncompressed signal. I have followed this pace very closely, as I want to buy a display with DP80. Real suckers in monitor space... From this point of view, I am not surprised that AMD decided not to upgrade DP54 on 9000 client cards, as display industry remained very sluggish in adopting the new standard and they charge fortune for it as if it's made of gold.

Now, they will probably be the first, again, to offer HDMI 80. I can guarantee you that monitor vendors will be sluggish again and TVs will beat them with new HDMI ports, preparing for new consoles and GPUs. So, tame your emotions and rushed judgments. The industry is a complicated mess.

Dr. DroThere's no way this is Nvidia's fault, not in the slightest - and with their Arc A770, Intel went as far as devising a workaround for HDMI's restrictions on open source by using a hardware signal converter (the PCON). The only company that deliberately imposed any artificial limitations here is AMD, and these pertain only to DisplayPort anyway.
Completely Nvidia's fault, as they chose not to develop a new PCB with faster ports for 4000 card when technology was available, to save on R&D costs, and push more people to upgrade sooner when they discover that debilitated display heads on DP 1.4a cannot drive more than two 4K/240 monitors with DSC, blocking the output on other two ports. Check the spec of 4090. You probably remember the embarrassing situation when 4090 was not able to drive Samsung's 8K/2K 57-inch monitor at 240Hz, but it managed only 120Hz image due to limited display heads. RDNA3 cards did not have this issue at all, but the then most expensive consumer card on the planet did. Read about it to find out more.

The use of PCON on some Intel Alchemist cards has nothing to do with HDMI Forum or open source. Intel's silicon does not natively support FRL signalling that ensures speeds above 18 Gbps signal provided by older TMDS signalling on HDMI 2.0b. Intel chose not to offer native support for HDMI 2.1 FRL on their cards at the time when both Nvidia and AMD offered HDMI 21. FRL on 3000 and 6000 cards.
Posted on Reply
#18
Rover4444
Dr. Drowith their Arc A770, Intel went as far as devising a workaround for HDMI's restrictions on open source by using a hardware signal converter (the PCON).
Yet another Intel W. Thank you, Intel.
Posted on Reply
#19
Dr. Dro
Tek-Check"It's not that hey did not want... " Where do you take such nonsense from?
If they had really wanted to introduce DP 2.1 in 2022, they had all tools to do it, like AMD and Intel. Nobody was hiding new ports from them. It was enough to invest into new PCB and install the standard. Instead, they were stingy and decided to recycle the PCB from 3000 cards, thus slowing down the adoption of DP 2.1 in the entire monitor industry for two years, apart from 2-3 models that came out in 2023, such as G95NC.
In this way, Nvidia showed a middle finger to all buyers of 4000 cards. Do not defend them for not using available technology.
I appreciate you taking the time to make a well thought response and a quality, justified post full of very valid information.

I'm not sure it's that simple, a GPU takes a very long time to develop, afaik the release cycle from floorplanning to releasing a new architecture is around 3 to 4, perhaps even 5 years of engineering and development time. VESA released the 2.0 specification in late 2019, just a few months before the launch of Ampere. At that point it's safe to say Ada was already in advanced stages of development and even some preliminary work already being done on Blackwell - and I'm quite sure that the groundwork for RDNA 3 and 4 was similarly just about starting at around that point, too. Blackwell is just out now and the Rubin architecture is all but completed (AI/datacenter Rubin Ultra GPUs are expected to ship next year), with the first details on Feynman architecture coming in tow, AMD's plans for the reunification of CDNA and RDNA emerging at the same time. If I had to say, AMD's probably already starting the preliminary work on UDNA 2 by now. I'm sure Nvidia could have rushed out a spec upgrade to get DP 2.0 implemented, but it'd probably result in a poor quality product with lots of bugs, and we've been seeing some even now with drivers and BIOS updates issued to the RTX 50 cards to support DP 2.1 properly.

Being limited by the industry is a relatively poor excuse, IMO. GPUs will stick around longer than displays and the proper tools to validate such usage scenarios exist, even if monitors aren't widely available yet. For the mini-DP stuff, I can't claim I have deep knowledge of how the electrical side works, but DP 2.1 on full size DP with 50 series has been done - reiterating what I just mentioned about bugs being present. Perhaps this is the reason why AMD has decided to "throttle" the mass market a tier below.

Good info on Arc, although that is certainly a nice benefit, PCON has been allowing these cards to do 4K/120 on Linux after all :p
Posted on Reply
#20
Tek-Check
dgianstefaniI just want AIB cards to start coming with a Type C output again, like they used to couple gens ago. Something like 2x DP, 1x HDMI, 2x Type C (DP).
My previous reference model of 7900XTX actually had USB-C with DP54. The problem is that vast majority of monitors do not have such offer.
Posted on Reply
#21
ymdhis
TheinsanegamerNA/V equipment often still has VGA on it. Not our fault they're so far behind.
We are talking about equipment that people use for decades. Supporting older standards can be a major selling point. And even so, I don't remember much A/V equipment with VGA ports, except maybe projectors and TVs. I'm not familiar with the wonderful world of video cameras though.
Bruno_Oplease do show me 1 single (2024+) AVR that has a VGA port.... everything is 100% HDMI based for years now
I can't even name 1 AVR that had VGA port, ever, let alone 2024+ models. Even the legendary flagship models like Denon A1, Yamaha DSP, or Onkyo Integra series (flagship models with bajillion inputs) never had those in any of its iterations going back 30 years.

edit: after some searching it looks like the highest-end Onkyo models between 2010-2014 had VGA ports. Huh.
Posted on Reply
#22
Tek-Check
Dr. DroI appreciate you taking the time to make a well thought response and a quality, justified post full of very valid information.
Thanks. Appreciate the feedback.
Dr. DroI'm not sure it's that simple, a GPU takes a very long time to develop, afaik the release cycle from floorplanning to releasing a new architecture is around 3 to 4, perhaps even 5 years of engineering and development time. VESA released the 2.0 specification in late 2019, just a few months before the launch of Ampere. At that point it's safe to say Ada was already in advanced stages of development and even some preliminary work already being done on Blackwell - and I'm quite sure that the groundwork for RDNA 3 and 4 was similarly just about starting at around that point, too. Blackwell is just out now and the Rubin architecture is all but completed (AI/datacenter Rubin Ultra GPUs are expected to ship next year), with the first details on Feynman architecture coming in tow, AMD's plans for the reunification of CDNA and RDNA emerging at the same time. If I had to say, AMD's probably already starting the preliminary work on UDNA 2 by now.
This has nothing to do with Nvidia betraying their buyers of 4000 cards when new display technology was simply available. Again, they chose not to offer it when AMD chose to offer it. VESA told them what would be available and what could be achieved with the then technology. GPU companies have choices and decisions to make.
Dr. DroI'm sure Nvidia could have rushed out a spec upgrade to get DP 2.0 implemented, but it'd probably result in a poor quality product with lots of bugs, and we've been seeing some even now with drivers and BIOS updates issued to the RTX 50 cards to support DP 2.1 properly.
They could have offered in 2022 the very same thing as AMD did, DP54 on consumer cards and DP80 on professional cards. RDNA3 cards did not have issues with video signal, at least nothing was widely reported. Problems on 5000 cards is a separate matter. Perhaps they could invest more in R&D stage to iron out issues before real buyers become frustrated beta testers.
Dr. DroBeing limited by the industry is a relatively poor excuse, IMO. GPUs will stick around longer than displays and the proper tools to validate such usage scenarios exist, even if monitors aren't widely available yet. For the mini-DP stuff, I can't claim I have deep knowledge of how the electrical side works, but DP 2.1 on full size DP with 50 series has been done - reiterating what I just mentioned about bugs being present. Perhaps this is the reason why AMD has decided to "throttle" the mass market a tier below.
You can't have it both ways, defending Nvidia for not offering new ports with any of three available UHBR speeds earlier and criticizing AMD for continuing to offer well working DP54 ports, while many Nvidia 5000 cards have issues with DP80 signal and black screens. It's not binary. It's complex. But hey, at least we have a leak about HDMI 2.2 on RDNA5, which is exciting.
Posted on Reply
#23
trsttte
Little is known of the HDMI 2.2 spec yet, but if they're "only" claiming 4k 240hz at 12bit that would be an effective bandwidth of 82.27 Gbit/s against DP 2.1 UHBR3 77.37 Gbit/s. A win is a win, but they're really grasping at straws here, especially when their protocol requires a 96 Gbit connection and for display port a 80 Gbit is enough. The cable requirements for DP2.1 are already hard enough, HDMI will be even worse for barely any gain

Why bother are you bothering AMD?! People won't play games with 12bit colors and they won't watch movies with 240Hz motion, the "advantages" of HDMI 2.2 don't really matter, HDMI 2.1 is enough in the places where HDMI matters, we have Display Port for the rest.
Posted on Reply
#24
Dr. Dro
trsttteWhy bother are you bothering AMD?!
Why bother? I'll tell you why. It's because this will enable things like 10 bpc 8K at 60 Hz, 8 bpc 4K at 360 Hz, 8 bpc 1080p at 1440(!) Hz. It's also the stepping stone to resolutions beyond 8K. There's a monitor revolution simply waiting on such a link standard to become widely adopted, and it'll be DP that will have to up their game to match it.
Posted on Reply
#25
Rover4444
trsttteLittle is known of the HDMI 2.2 spec yet, but if they're "only" claiming 4k 240hz at 12bit that would be an effective bandwidth of 82.27 Gbit/s against DP 2.1 UHBR3 77.37 Gbit/s. A win is a win, but they're really grasping at straws here, especially when their protocol requires a 96 Gbit connection and for display port a 80 Gbit is enough. The cable requirements for DP2.1 are already hard enough, HDMI will be even worse for barely any gain

Why bother are you bothering AMD?! People won't play games with 12bit colors and they won't watch movies with 240Hz motion, the "advantages" of HDMI 2.2 don't really matter, HDMI 2.1 is enough in the places where HDMI matters, we have Display Port for the rest.
AMD's bothering because HDMI won't let 2.1 be implemented under Mesa so they're leapfrogging to 2.2 so they can implement it before HDMI says no this time :pimp:
Posted on Reply
Add your own comment
Nov 21st, 2026 23:22 PST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts