superMHL: reaching for 8k at 120Hz

[Black Octagon]

I know the range of connectors we have today is already too much, but (drooooooool....):
http://www.anandtech.com/show/8843/mhl- ... rive-8k-tv

If I've been following things correctly, we now just need the panels to catch up
(I'm looking at YOU, OLED).

[Edmond]

[jts888]

Sounded pretty good until I saw the 8k@120Hz w/ 4:2:0 chroma subsampling.
I'd like an 8k monitor someday, and subsampling makes vector elements including text look like garbage.
Could be great for sports television, etc., though if the cable companies bother relay sufficient bandwidth data streams.

[flood]

usually its the scaling of the chroma layers which make sharp lines look like shit. with good scaling it will just look a bit soft

[Edmond]

Sounded pretty good until I saw the 8k@120Hz w/ 4:2:0 chroma subsampling.
I'd like an 8k monitor someday, and subsampling makes vector elements including text look like garbage.
Could be great for sports television, etc., though if the cable companies bother relay sufficient bandwidth data streams.

Isnt 4:2:0 half the bandwidth use of 4:4:4 ?
So with 4:4:4 the superMHL cable should handle 8k@60hz?
And its still 36bit color depth.

I dunno. Seems that way to me that the bandwidth of this cable is at least 70.00 gbit/s.

Also, think of the applications for other resolutions.
Like 2000x1000 @ 1000hz with 30bit color depth is 60 gbit/s, or 4000x2000@ 250hz with 30bit color depth is also 60 gbit/s.

[jts888]

I dunno. Seems that way to me that the bandwidth of this cable is at least 70.00 gbit/s.

Yeah, my math checked out the same, and I got pretty skeptical.

Even super-expensive QSFP copper cables with heavily shielded twinax (differential pairs inside dedicated shielded sub-channels) struggle to get more than 13 Gbps per lane, and the minimum PC host bus needed to reach those rates would be a 16x PCIe 3.0 link.

Even if they could squeeze 12 usable signal lanes out of the 32 pins, getting 6+Gbit per lane to work reliably on consumer-grade cabling with an error intolerant protocol (HDCP) is almost completely absurd.

This technology involves either additional undisclosed compression, magic, or lies.

[flood]

yup im calling bs

[jts888]

FWIW, I looked into MHL, and version 3.0 from last spring supposedly does 6 Gbps, and superMHL is supposed to have 6 times as many lanes with each clocked at double the rate, actually hitting 72+Gbps maybe.

The older systems worked over USB 3 connectors and so were probably using both high speed data lanes for data transmission at 3 Gbps per pair.

superMHL may actually be running 12 pairs at 6Gbps each here somehow.

[Edmond]

This technology involves either additional undisclosed compression, magic, or lies.

Maybe im super optimistic but i dont see why they couldnt had just made a really fat copper cable...
Imagine if vesa made DP1.4 that is just two DP1.3s stuck in one cable with a single connector at the end instead of two.

Or am i not getting this at all?

[Black Octagon]

I would like to say that I don't appreciate haters entering my thread and turning my hard-on into a floppy disappointment

If we can crack the human genome and send David Bowie fans to the moon, we can surely make this work somehow. God knows these companies are rich enough and eager enough for even more of our cash.