Newer! Faster! Brighter! (And Barely Fast Enough)
Back in September, we discussed the escalation of “Ks,” as in how many thousands of pixels the display industry is trying to stuff into next−generation LCD, OLED, and inorganic LED panels. We mentioned that the first 8K displays are now coming to market, even as our industry is still trying to come to grips with the care, feeding, and handling of 4K / Ultra HD video signals.
Things are moving more quickly than anticipated. The HDMI Forum recently held a press conference in New York City to talk about HDMI 2.1 and where it’s headed. This newer, faster version of HDMI was first introduced at CES in 2017 and is quite the departure from previous versions.
Instead of using transition−minimized differential signaling (TMDS), which was the foundation of digital display interfaces going back to DVI in 1999, version 2.1 has adopted a packet format very similar to that of DisplayPort. By doing so, HDMI 2.1 can now expand signal carriage to four lanes of data with an embedded clock, compared to the older 3 lanes with separate clock used in all HDMI versions through 2.0.
There are other advantages. Because the signal is 100% digital now, it can be compressed using Display Stream Compression (DSC), which will come in really handy with the massive signals needed to handle high frame rate video and 8K. Another advantage is that the clock rates and data are free to zoom far beyond the 18 Gb/s limit of version 2.0.
Indeed; HDMI 2.1 now has a maximum data rate of 48 Gb/s (or 12 Gb/s per lane). That number is mind−boggling: We’re only starting to see network switches with that much speed come to market. But if you run the numbers, you WILL need that kind of speed for advanced high−resolution imaging.
Consider a 4K signal with high dynamic range and a 120 Hz frame rate. The base clock rate for such a signal, using standard CTA blanking, would be 4400 pixels (x) 2250 pixels (x) 120, or 1188 MHz (1.188 GHz. Add in 10−bit color (the minimum for HDR) with 4:4:4 (RGB) color resolution, and the grand total (after shopper coupons) is 1188 (x) 12 (x) 3 = 42.77 Gb/s. Going to lower color resolution lowers the tab a little: With 4:2:2 color, the data rate is 28.51 Gb/s and with 4:2:0 color, it drops to 21.39 Gb/s.
That’s still pretty fast – too fast for HDMI 2.0. And if we start talking about 8K imaging, things get even crazier. An 8K video stream (again, using standard CTA blanking) with just 10−bit RGB color at 60 Hz refresh will leave you in a cloud of dust:
8800 (x) 4500 (x) 60 (x) 12 (x) 3 = 85.536 Gb/s.
Zoom−zoom! We’d have to drop to 4:2:0 color resolution just to get that signal through an HDMI 2.1 connection. Even 4:2:2 color would be too fast at about 57 Gb/s. The current version of DisplayPort would also vanish in the rear−view mirror, as it is capped at 32.4 Gb/s. (We expect to hear about a new version of DP at CES next month, presumably one that’s a LOT faster.)
This is presumably where DSC would enter the picture. It is capable of 2:1 compression with extremely low latency, and that would get our example 8K/60 signal down to earth and to a point where it could travel over HDMI 2.1 (but not DP). The only catch is, DSC requires quite a bit of computation to work correctly and is considered “CPU−hungry,” which of course adds cost to its implementation.
What’s curious about HDMI 2.1 to us is the continued lack of a native optical transport specification. Any signal running in the 40 Gb/s range should probably travel over optical fiber. Certainly, if it’s going to travel through a 40 Gb/s network switch, that transport will be as pulses of light and not electrons dancing on the outer edge of copper conductors.
We inquired at the NYC press event if any HDMI Forum members were actually making v2.1 transmitter and receiver chipsets yet. So far, only one company in Japan (Socionext) is doing that, but you would be hard−pressed to find any commercial or consumer products that support V2.1 at present. (We’ll certainly have our eyes open at CES for one!)
As mentioned in September, it’s expected that over 5 million 8K TVs will be shipped worldwide by the end of 2020 – just two years from now. Hand−in−hand will be a small but growing number of 8K monitors for commercial use (yes, there are customers waiting for such products, believe it or not) and the vast majority of those will come from super−sized LCD panel “Fabs” in China that are currently under construction or just firing up.
We’ve frequently used this expression in the past: “What good is a Ferrari if you live on a dirt road−” Well, that’s pretty much the situation we’re looking at with the next generation of displays. Higher resolution, high dynamic range, wider color gamuts, and high frame rates will all add up to super−sized packages of display data that dwarf what we switch and distribute today.
New codecs like JPEG XS / TiCo will help to squeeze things through network switches, but we’ll still have a choke point at the physical display interface. And we don’t have any real solutions to the problem just yet: Do we use compression− Double up on interface connections− Skip the traditional HDMI / DP interface altogether, and use a decoder inside the display to decompress the signal−
FIND WHAT YOU NEED
Want to receive alerts and updates for every new post?