HDMI 2.1 Display Stream Compression (DSC) Ready for Amdgpu Linux Driver | Hacker News Reader

HDMI 2.1 Display Stream Compression (DSC) Ready for Amdgpu Linux Driver

phoronix.com

66 pointsWithinReason49 comments

Every time I hear about HDMI and amount of legalese problems around it, my response is "you can just use DisplayPort"

Most TVs don't have DisplayPort

In my case its an issue because I have a monitor with only a single DP port and I need to switch between my tower and laptop. I have to use HDMI for the laptop to monitor connection.

You can get DisplayPort KVMs. As a nice bonus the KVM will let you share a single mouse and keyboard set between them.

My monitor (Samsung Odyssey Neo G9) has a usb kvm built in, i can already do that.

Plus I havent really seen an external dp2.1 kvm switch yet and I'm sure if they exist they are expansive.

An adaptor costs £7.

They are not equivalent.

Conversion is a very intricate spec fulfilment over an incredibly high bandwidth signal.

I did the dive; The adapters are not sufficient.

Do you know whether HDMI CEC adapters impact the signal?

In what way are they not sufficient?

The Synaptics VMM7100-based adapters only support VRR on older firmware versions with bugs.

The Chrontel CH7218 is the most reliable but still also suffers blackouts during VRR.

ParadeTech PS196 adapters advertise VRR support but their DPCD does not correctly communicate that it is supported. So even if you add the chip to the VRR PCON list in the amdgpu driver, it still won't see VRR as supported.

And while some of these advertise themselves as displayport 2.0, all of them only support bandwidth of 25.96gbps on the displayport side, requiring DSC for 4k 120hz 10bit color, even though they support 48gbps on the HDMI output.

For HDMI 2.0. For HDMI 2.1 and 4K/120hz you're looking at north of $25 and don't get VRR support.

Fair enough about VRR, but £13 for 4k/240Hz - https://www.amazon.co.uk/dp/B0FQCF62CD

I've owned 2 of these (returned and reordered thinking the first might've just been bad) and neither worked properly on Linux with an AMD 9070xt and an LG CX. They'd have black screen dropouts every few minutes, and occasionally full screen color corruption.

with lossy compression

They have limitations, specially when driven to the limits of the specifications.

When doing 4k@120fps 4:4:4 chroma you might have to deal with longer handshakes and sometimes even no handshake at all. Or random dropouts. Or HDR not activating properly.

I thought handshakes were just when you were setting up a connection no?

Random dropouts tho sound bad… with high speed signaling also sounds like a pain to figure out

There are only a few adapters that support the 2.1 features (hdr+vrr+high resolution+high refresh rate, no lossy DSC). I even had to flash custom firmware for most of those features to work (vrr still doesnt)

But wouldn't this break the HDCP chain and therefore render many use-cases (playback of DRM-protected streams) broken?

Is that a problem for most uses of DP?

Can DisplayPort be used for audio? My recollection is that the main advantage of HDMI is transmitting both video and audio.

See "Audio specifications" at bottom of table:

* https://en.wikipedia.org/wiki/DisplayPort#Specifications

* https://www.cablematters.com/Blog/DisplayPort/does-displaypo...

You can just use Nvidia.

no, Nvidia is the HDMI of gpus

This was previously blocked from inclusion in SteamOS by the HDMI forum. It would help the Steam Machine to each 4K120Hz on HDMI.

It was blocked from inclusion in the AMD GPU drivers, it's nothing specific to Steam or the Steam Machine.

The HDMI Forum apparently forbids any open source implementation of HDMI 2.1. Although I don't know if they ever offered an official justification, for a group that exists to promote HDMI adoption, they're clearly morons.

It's a group that exists to make sure that the standard works for all the members, including media companies that think they can control the flow of information. They don't need to promote HDMI adoption since their members already control pretty much all the TV production.

It's still not clear why the standard wouldn't work by AMD having an open source implementation. I try to give the benefit of the doubt, but in this case, it's hard not to agree with the parent comment: whoever came up with this rule and the people who agreed with it are morons.

were*

As the article says, they most likely changed their mind, probably following quite a bit of background discussions and industry influence.

Wait "forbids", present tense? How does that track with this announcement that it's coming to the open source AMDGPU?

This is actually a good question … wonder if it’s just api calls to a binary blob haha

If memory serves HDMI includes DRM which they don't want people to reverse engineer.

Valve also contributes to the amdgpu driver

> 4K @ 240Hz

WHY!?

I have a Samsung G95NC (DP2.1, 7680x2160, 240Hz) and you definitely notice the difference between 120Hz and 240Hz. Although personally I wouldn't pay a cent more for an even higher refresh rate since the difference is much less noticable than 60vs120Hz and I expect 240Hz vs 420 to make even less of a difference.

Would you rather they explicitly blocked that even though the technology allows for it?

It's too easy for display manufacturers to compete on moar pixels, moar fps, moar refresh. You just try to embiggen your numbers compared to your competitor.

Meanwhile, features where you can't compete on numbers but can ruin the experience are ignored.

What are these features that can't be measured?

Plenty of people who test monitors also compare things like color coverage, brightness, latency, contrast, viewing angles, etc, etc, etc. If you mean the entire monitor, they generally also cover things like how the display swivels/mounts among other things.

Brightness and contrast are also gamed to hell and back. You have to go to some very specialized review sites to read about contrast at low brightness (lights off software development scenario, or maybe even lights off gaming).

I must admit I've sometimes seen a description of how wobbly the monitor is when you hit your mechanical keyboard like it's stolen your car and killed your dog. Veeery rarely though.

No, I just don't think such a high refresh rate accomplishes anything. Not even bragging rights. 120Hz, possibly. But 240? Are you going to introduce a telly into a slow-motion studio, on the set?

The difficult thing for these standards is the data rate. 4k 240Hz is the same data rate as 8k 60Hz (since 8k is 4x the number of pixels as 4k).

If you want to support 8k 60Hz, the only reason you wouldn't also support 4k 240Hz would be because you actively choose to disallow that. That seems like a bad idea.

I guess that would sorta make sense if there was no upscaling involved. With upscaling, I doubt the actual display refresh rate ever hits 240Hz. They could still advertise support for it, but you'd have to use a high speed camera to tell if it really does that.

I have no idea what you mean. Why would upscaling make it so that a display can't hit 240Hz?

Because the amount of pixels on a flat display is fixed? So 8K is always 8K and if it does 8K@120, the display is really just 120Hz, no?

Yes, 8k @ 120Hz is 120Hz. Did I ever imply anything else? And who brought up 8k @ 120Hz?

I've been talking about 8k @ 60Hz and 4k @ 240Hz, because those are the same data rate; 1 990 656 000 pixels per second. One expressed as 7680 x 4320 x 60, the other expressed as 3840 x 2160 x 240.

What does any of this have to do with upscaling?

Because an 8K pixel display is always 8K. So the display on a typical TV always has the same mode. That could mean your 4K@120 signal could (in theory) translate to 8K@60 internally. The data rate over HDMI would still be the same. HDMI is just an interface. What the TV does internally is up to the manufacturer.

Besides, it wouldn't be the first time a TV advertises a higher mode than the display can actually handle.

Who mentioned 4k@120?

How could 4k@120 be translated to 8k@60 internally? 8k@60 is twice the number of pixels per second as 4k@120.

I still have no idea what you're trying to say.

You're either being overly pedantic or are just refusing to acknowledge the point I'm trying to make. I'll just not bother anymore because it's a waste of time.

Brother I don't know what point you're trying to make, I don't understand what you mean by upscaling in this context and what some internal format conversion is meant to accomplish or why one would want that. What does internally converting 4k @ 240 to 8k @ 60 accomplish that just outputting 8k @ 60 directly from the computer wouldn't accomplish?

It's diminishing returns, but for those who want it and have the hardware to support it, why not?

I have a 360 and when I play something I can actually get a full 360 out of, it's wonderful! Though honestly anything over 100 I'm perfectly fine with.

Reply on news.ycombinator.com