superMHL: reaching for 8k at 120Hz

[Chief Blur Buster]

Displays and their bandwidth have their equivalent of Moore's Law, even if slower than for CPUs.

We're in an era where smartphones can record 4K video, y'know.

[Chief Blur Buster]

Blogged about this now.

Took pictures!

http://www.blurbusters.com/8k-120hz-vid ... -supermhl/

[masterotaku]

Good. Now we need monitors and GPUs with that connection. I'm already tired of seeing 4k 60Hz monitors without any type of strobing.

[Edmond]

Just wanna say that the superMHL 8k @ 120hz @ 36bit is also @ 4:2:0 palette.

If im not mistaken 4:2:0 is half the bandwidth of 4:4:4.

So according to my calculations superMHL would drive 4:4:4 @ 36bit @ 8k @ 60hz, MAX.
Or 71,66 Gbit/s
I bet the actual cable is 72 Gbit/s MAX tho.

have fun with seeing what kind of screens you could make with this ^^
https://emsai.net/projects/widescreen/bandwidth/

[Black Octagon]

Thanks for the link. Do we really need 36-bit for things like gaming, though?

[Black Octagon]

By the way, that site erroneously claims that Dual Link DVI's pixel clock is limited to 330Mhz

[RLBURNSIDE]

Thanks for the link. Do we really need 36-bit for things like gaming, though?

Yes we do. If we want to increase quality, increasing the pixel density has diminishing returns (except for VR), and increasing bit depth from 10 to 12 bits per channel results in a huge improvement in the ability to render wider color spaces, high dynamic range, and so on. HDR and wider color are probably the most important aspects of the new UHD Bluray spec, and TVs that implement HDR tech like Dolby Vision are going to blow people away in videogames. Yes, it is that impressive. A guy here at the office (I make games), saw the Dolby Vision 12-bit 4000 nit spec in an Unreal 4 demo and said it was the nicest looking tech he'd ever seen. It is going to be leaps and bounds more important to increase from SDR to HDR and wider color depth than it is to support higher def. Thankfully the current gen consoles can support 10-bit color and HDR no problem. Deep Color has been possible to send over HDMI since v 1.3. Finally there is a use for it.

[spacediver]

agreed, color bit depth, dynamic range, and color gamut are far more important to overall image quality than # of bloody pixels.

[flood]

i'd rather have high refresh rates (like >200) and framerate than any of those

[RLBURNSIDE]

i'd rather have high refresh rates (like >200) and framerate than any of those

No reason to not improve both IQ and FPS (to a point). But in terms of allocation of effort? No, super high FPS will not get any kind of priority from studios.

HDR and wider gamut increase rasterization costs by 25% or so (8 bit to 10 bit color depth), but asking for 200fps is kind of silly. As a game developer, it will be a long time before we optimize a AAA title for super high frame rate over things like draw distance, geometry density, AI, etc. I mean, PS4 and Xbox one have a hard time reaching 1080p60, and if you're talking about PC, it's still a pipe dream to aim for ultra high frame rate all the time. If you think going beyond 100fps as a development goal is more important than what each frame looks like, in terms of development resources and focus, I've got news for you, it's not going to happen. Even as GPUs increase, the target minimum frame rate on PCs will remain 60hz for a long time simply because that's what the majority of TVs are limited to. It's simple economics. PC monitors can go up to 144hz, but I'd MUCH rather play games on a 55 inch TV (or a projector) than a 24 or 27 inch monitor, and so would the vast majority of gamers out there. How do I know this? Because I work in the industry and have made dozens of AAA games, and the living room is where the money is. Basically, consoles.

I should also note to the admin, that asking for 1000fps to reduce blur to 0 is even more a pipe dream, it's never going to happen. Any increase in GPU power is going to be matched by a proportional increase in rendering workload. I do rendering and there's no way we'd waste our optimization efforts to target > 100fps instead of vastly increasing particle count, geometry density, draw distance, # of objects displayed. It's just not going to happen. And even if VR targets 120hz, any increases in GPU power beyond the ability to do that, won't be used to target 240fps instead, it will be used to increase detail instead. Reducing motion blur and lag are very important to VR, but people still want to actually play games on these things, and when game A is running at 480hz per eye with Counter-strike quality graphics and the other one is running something like GTA V at 120fps, which do you think players will play more? Or more importantly, spend more money on (I know CS is super popular, but it's archaic) and therefore get AAA development budgets to focus on.

It's diminishing returns going beyond 120hz, let's at least get to that point first before we start talking about 200fps+ let alone 1000fps rendering which is just not gonna happen for anything but the most basic graphics.

As an engineer, I put effort where it yields the greatest dividends. Corporations do the same. AAA studios do the same. Indie studios do the same. They go where the money is. There simply isn't enough demand for a game with simpler graphics but ultra high frame rate compared to more complex imagery with a still-decent 60p. Targetting beyond that is diminishing returns, both from a technical perspective, and definitely from a market and engineering perspective. Trust me, HDR is going to make the next gen games look stunning. Mark my words. We're working on it.