Time for optical video cables?

Quartz-1

Supreme [H]ardness
Joined
May 20, 2011
Messages
4,257
Has the time for copper video cables passed? How can we transmit the ever-higher bandwidths required for displays with higher and higher resolutions and frame rates? Will optical cables fit the bill?
 
Video Cables have not pass yet but yes on higher data rates possible with fiber optics, longer distances and the cable part can be cheaper. Cost would come in the transmitter and receiver to transmit and receive the information each way. I don't see that happening yet but maybe 10 years you never know.
 
Optical would certainly get around the length constraints with copper, and I don't think people will tolerate cables that need to be shorter than 3m for best signal quality. However, the industry is so entrenched in HDMI that I don't think we'll see any major changes anytime soon. Plus, moving away from copper will eliminate the passive compatibility with DVI, which is the Windows XP of the display world. But sacrifice is often required to move forward.
 
I'd be more interested in seeing optical traces going from the CPU to video card.

Don't kid yourself though, optical video out won't do anything for you. They aren't going to develop anything that is going to have all the bandwidth you'll ever need. Whether it's optical, HDMI, display port, they will limit it and create "versions" so they can continue selling new products with new/better features.
 
Optical would be way too expensive. You could add what, 500$+ to the card and monitor each not to mention the other downsides.
 
Optical would be way too expensive. You could add what, 500$+ to the card and monitor each not to mention the other downsides.

Optical monitor cables are expensive now, but ISTM that the expense is in the niche market and the fact that there are converters at either end. Fibre cable itself is very cheap, and optical NICs are nothing new. Apply economies of scale and the cost of the cable in its current form should come right down. Give the PC an optical out and the monitor an optical in and you should be laughing. Except it hasn't happened, so there must be a good reason, right?
 
Optical monitor cables are expensive now, but ISTM that the expense is in the niche market and the fact that there are converters at either end. Fibre cable itself is very cheap, and optical NICs are nothing new. Apply economies of scale and the cost of the cable in its current form should come right down. Give the PC an optical out and the monitor an optical in and you should be laughing. Except it hasn't happened, so there must be a good reason, right?


Then we get into the "why?" zone. The fact of the matter is as long as the needs can be met by cheap, passive cables at distances of 10ft or less, you have satisfied 99% of use cases. Those of us that want longer runs are an extreme minority.
 
I think the future belongs to "wireless" technologies because A) it sounds good to general populace (look ma, no wires) because nobody likes wires B) it is easy to market C) I would love to bybass drilling walls/windows/floors to take my PC to another room in order to avoid noise (it is so *very* quiet around where I live, and any noise is very easy to hear). Oh the never ending frustration until I found 4k/60Hz 5m (16 ft.) displayport cable which worked reliably (or for that matter USB 3.0 cable) because my pc is 5m away from my desk and I would love to get it farther but displayport gets very "snowflaky" over long distances.
 
I think the future belongs to "wireless" technologies because A) it sounds good to general populace (look ma, no wires) because nobody likes wires B) it is easy to market C) I would love to bybass drilling walls/windows/floors to take my PC to another room in order to avoid noise (it is so *very* quiet around where I live, and any noise is very easy to hear). Oh the never ending frustration until I found 4k/60Hz 5m (16 ft.) displayport cable which worked reliably (or for that matter USB 3.0 cable) because my pc is 5m away from my desk and I would love to get it farther but displayport gets very "snowflaky" over long distances.

And what do you think wirelesss will do to fix this? Wifi/Wigig is limited by maximum bandwidth available in crowded spectrum, and is stil about a quarter the bandwidth we need for these next-generation displays.

And LED can do this pretty well, but is point-to-point (reflections are under 100Mbit), and nobody has done it outside a lab, so again not cheap!

https://en.wikipedia.org/wiki/Li-Fi

And fibre has already been discussed - cheap CABLE, but never as cheap as copper, because of the TWO optical transceivers each completed connection requires. Thats several hundred dollars for something fast enough to handle 40Gbps!

See these 16Gbit/s SFP+ modules, which are not fast enough to replace the new HDMI.

https://www.amazon.com/Cisco-SFP-10G-LR-10GBASE-LR-SFP-Module/dp/B0026BQHSA

So copper remains our only option, because it uses the same native electrical signals as the CMOS processors that manage our data, and it's way faster and more reliable than wireless.
 
Last edited:
As an Amazon Associate, HardForum may earn from qualifying purchases.
Optical monitor cables are expensive now, but ISTM that the expense is in the niche market and the fact that there are converters at either end. Fibre cable itself is very cheap, and optical NICs are nothing new. Apply economies of scale and the cost of the cable in its current form should come right down. Give the PC an optical out and the monitor an optical in and you should be laughing. Except it hasn't happened, so there must be a good reason, right?

What does a 10Gbit fiber NIC cost with it all?

If you watched Intel when they talk about photonics for example in the DCG sessions. You see optical cost is one of the biggest burdens to datacenter infrastructure.

A dual 10Gbit fiber NIC cost ~1000$ with all you need and its not because its a low volume product. Now imagine something like 100Gbit with a connector or 2 in monitors and 3-4 connectors on a graphics card.
Even if you get the cost down to 200$ per connector, we still talk close to 1000$ for monitor+graphics card in connectors alone.

Also remember optical is anything but trouble free. It cant handle what electrical can in terms of daily punishment and abuse.
 
Last edited:
As I understand it, fibre cable bandwidth is affected by the length of the cable due to dispersion and attenuation. If you know you've got a short cable, you can use more bandwidth. Remember that standard fibre is expecting to operate at over 1 km. But if you have a length limit of 10m then you can push petabits.
 
As I understand it, fibre cable bandwidth is affected by the length of the cable due to dispersion and attenuation. If you know you've got a short cable, you can use more bandwidth. Remember that standard fibre is expecting to operate at over 1 km. But if you have a length limit of 10m then you can push petabits.

That's not true at all. SX depending on cable is 2 meters to 550 meters, LX can handle up to 5km and LH up to 10km. Beyond those you use different methods.

The most common used setup is SX and a 62.5um multimode fiber with around 250 meters distance as max.
 
A dual 10Gbit fiber NIC cost ~1000$ with all you need and its not because its a low volume product.

But is it a high volume product? Or is it stuck in the netherworld between niche and mass market? I'm completely blanking on sales figures, but I do note that per port pricing is coming down rapidly. If you start banging out tens and hundreds of millions of these things - every motherboard, every video card, every monitor - economies of scale kick in and the price drops precipitously. Remember how expensive gigabit networking used to be? Now it's the home standard. 10G will follow.

We already have a standard for fibre in TOSLink, but the bandwidth for that is only 125 Mb/s, so the start is there.
 
But is it a high volume product? Or is it stuck in the netherworld between niche and mass market? I'm completely blanking on sales figures, but I do note that per port pricing is coming down rapidly. If you start banging out tens and hundreds of millions of these things - every motherboard, every video card, every monitor - economies of scale kick in and the price drops precipitously. Remember how expensive gigabit networking used to be? Now it's the home standard. 10G will follow.

We already have a standard for fibre in TOSLink, but the bandwidth for that is only 125 Mb/s, so the start is there.

Its a mass volume product through a really long time.

Even 10Gbit copper is so expensive still that 2.5 and 5Gbit standards have been created.

There is a long way between Toslink and ~40Gbit. And Toslink is ~35 years old and started at 3Mbit. Maybe in 10 or 20 years when photonics becomes a mass consumer product it can be done. But it still doesn't solve the issues by using fiber in the consumer space.

http://www.intel.com/content/www/us...con-photonics/silicon-photonics-overview.html
 
Last edited:
Both have their uses, but I doubt we see optical widely deployed beyond longer distances or sensitive systems. Big point to keep in mind, copper works better than fiber for carrying power which can be a good and bad thing. Thunderbolt for example powering a display and providing signal is a nice benefit. Fiber to your house or as a means of electrical isolation also has it's uses. Part of why fiber works well in a datacenter is avoiding current loops and EM interference everywhere. Surges where something fails and proceeds to kill everything electrically connected to it.
 
I figured by now the interconnects in a motherboard would be optical considering the speed increases and since it such short distances the cost would not be as high as an external cable.
 
Optical as a universal connector probably won't become mainstream until compute components also become purely optical. For now, all electrical to optical connections have to be converted (on both ends), adding to latency.
 
Even 10Gbit copper is so expensive still that 2.5 and 5Gbit standards have been created.

I'de be careful with assertions like that. Most nBaseT ports will end up being so close in hardware to 10GBaseT that we may not see it appear in many applications other than full range ports (10/100/1000/2500/5000/10000) on anything other than APs and consumer dumb switches.

2.5/5.0 is pretty much the 10G PHY, with less stringent modulation rates/bandwidth to allow use on Cat5E and Cat6. Cat5E use is the real killer app for it. No one wants to replace the existing Cat5E infrastructure, which is partly why 10GBaseT has been DOA (that and price + power + heat).
 
I figured by now the interconnects in a motherboard would be optical considering the speed increases and since it such short distances the cost would not be as high as an external cable.

There isn't any speed benefit using optical over electrics. Electrics actually tends to be faster. Opticals obviously got other benefits.
 
I think people are too focused on Fiber networks, for data center use, you don't want thick cables that are hard to bend, so you'll see dual port connections (2 cables) for a monitor cable, you can easily stick 4-8 cables in a bunch, and still still probably get enough flexibility for use, though I don't think there is a need for it, since 99% of monitor cables are short, and can carry enough bandwidth for what we are using.
 
My interest isn't networks but video bandwidth. Copper isn't coping over any but the shortest lengths so we need a replacement.
 
I'm still using DAC instead of fiber whenever the runs will allow.

Copper is going nowhere.
 
I think people are too focused on Fiber networks, for data center use, you don't want thick cables that are hard to bend, so you'll see dual port connections (2 cables) for a monitor cable, you can easily stick 4-8 cables in a bunch, and still still probably get enough flexibility for use, though I don't think there is a need for it, since 99% of monitor cables are short, and can carry enough bandwidth for what we are using.
Hard to bend? Do you mean not bendable? No one should be bending any fiber unless you want a dead cable.
 
Something needs to be done about DP and HDMI cable lengths. I have had serious issues trying to get 4K 4:4:4 running with my TV at 7+ meters lengths, even High Speed active cables have been unable to do it, either by not showing 4:4:4 option at all, flickering and even dropping refresh rates at lower resolutions. I probably have to buy some expensive cable to be able to do this. DP is generally even worse when it comes to cable length.
 
I think people are too focused on Fiber networks, for data center use, you don't want thick cables that are hard to bend, so you'll see dual port connections (2 cables) for a monitor cable, you can easily stick 4-8 cables in a bunch, and still still probably get enough flexibility for use, though I don't think there is a need for it, since 99% of monitor cables are short, and can carry enough bandwidth for what we are using.

Hard to bend? Do you mean not bendable? No one should be bending any fiber unless you want a dead cable.

Tell that to Samsung? Their new Premium line of QLED sets (and maybe non QLED too?) which have that breakout box, will now be connected to the TV via a --seemingly, by the looks-- single fiber cable. And on top of that, check out the bend they demonstrate...
samsung-qled-tv.jpg

"All of the Q sets use Samsung's One Connect box to house inputs, which, new for 2017, connects to the TV via a 5 meter fiber-optic umbilical that's nearly invisible. A 15-meter option will also be available. The One Connect box can also blast out IR commands to control cable boxes and other devices, to more easily stow your components inside a cabinet."

Something I'll be the first to admit, though... This is CES, this is a product demonstration, and no doubt a prototype.
That all means that what is shown is possibly just a mockup and non-functional, and/or not representative of what will come to market.
(I was only linked that image and text, and so I don't know what the original article stated on the tech...)
 
Tell that to Samsung? Their new Premium line of QLED sets (and maybe non QLED too?) which have that breakout box, will now be connected to the TV via a --seemingly, by the looks-- single fiber cable. And on top of that, check out the bend they demonstrate...
samsung-qled-tv.jpg

"All of the Q sets use Samsung's One Connect box to house inputs, which, new for 2017, connects to the TV via a 5 meter fiber-optic umbilical that's nearly invisible. A 15-meter option will also be available. The One Connect box can also blast out IR commands to control cable boxes and other devices, to more easily stow your components inside a cabinet."

Something I'll be the first to admit, though... This is CES, this is a product demonstration, and no doubt a prototype.
That all means that what is shown is possibly just a mockup and non-functional, and/or not representative of what will come to market.
(I was only linked that image and text, and so I don't know what the original article stated on the tech...)
True but i didn't even see that damn that's crazy.
 
Hard to bend? Do you mean not bendable? No one should be bending any fiber unless you want a dead cable.

Umm, not sure what you're talking about :) I'd send you a picture of our lab, but I'm sure I'd be breaking NDA :(

I'm no talking about sharp 90 degree bends, just bending in general.
 
Last edited:
Tell that to Samsung? Their new Premium line of QLED sets (and maybe non QLED too?) which have that breakout box, will now be connected to the TV via a --seemingly, by the looks-- single fiber cable. And on top of that, check out the bend they demonstrate...
samsung-qled-tv.jpg

"All of the Q sets use Samsung's One Connect box to house inputs, which, new for 2017, connects to the TV via a 5 meter fiber-optic umbilical that's nearly invisible. A 15-meter option will also be available. The One Connect box can also blast out IR commands to control cable boxes and other devices, to more easily stow your components inside a cabinet."

Something I'll be the first to admit, though... This is CES, this is a product demonstration, and no doubt a prototype.
That all means that what is shown is possibly just a mockup and non-functional, and/or not representative of what will come to market.
(I was only linked that image and text, and so I don't know what the original article stated on the tech...)

Far from reality, and where is the power cable. ;)
 
Umm, not sure what you're talking about :) I'd send you a picture of our lab, but I'm sure I'd be breaking NDA :(

I'm no talking about sharp 90 degree bends, just bending in general.
Slight curving yes. Straight bends no. Not at our facility.
 
Something needs to be done about DP and HDMI cable lengths. I have had serious issues trying to get 4K 4:4:4 running with my TV at 7+ meters lengths, even High Speed active cables have been unable to do it, either by not showing 4:4:4 option at all, flickering and even dropping refresh rates at lower resolutions. I probably have to buy some expensive cable to be able to do this. DP is generally even worse when it comes to cable length.
Expensive cables will give you no greater benefit over a long run. This is the reality with high bandwidth over copper. You simply need to move your video source closer to the display.

You think DisplayPort is bad, wait until HDMI 2.1 hits the market. Maximum bandwidth will only be guaranteed up to two meters.
 
To everyone saying that Optical automatically increases cost dramatically in Devices?

Remember: TOSLINK is a thing and it's on 10$ cards anymore.
 
There is certainly fiber cable flexible enough to wrap tightly around your index finger. I see them in patch cords frequently. But the electronics to utilize the fiber will likely be more expensive than copper, at least initially. After all, we will want 0 latency, unlimited bandwidth and color depth, etc. :)
 
Sure it is, but it's optical and a standard. Saying to converting to optical will be price prohibitive is not borne out by the industry.
Would it be cheap immediately? No. But neither were DVI Monitors 20 years ago.

Implementation, as it becomes common, would become cheap. (As long as Apple or Sony isn't the one driving it.)
 
Sure it is, but it's optical and a standard. Saying to converting to optical will be price prohibitive is not borne out by the industry.
Would it be cheap immediately? No. But neither were DVI Monitors 20 years ago.

Implementation, as it becomes common, would become cheap. (As long as Apple or Sony isn't the one driving it.)

Take a look on how long it took for 100mbit, 1gbit and 10gbit Ethernet to come down in price. And that's just a factor 100x. Then look on the price of fiber for the same.
 
Take a look on how long it took for 100mbit, 1gbit and 10gbit Ethernet to come down in price.

There's a point of demand to be considered. That demand can be created in some markets but not others. I believe that consumer video cables - and, by extension :), computer monitor video cables - are one such market.
 
If you don't want to do any R&D and don't mind a rather substantial QSFP plug, there is already an infiniband standard that can push 200Gbps (66/64b encoding) over a beefy copper DAC for a few meters. Fiber transceivers available as well for whatever range you want, and you can move any data not just "video".

Of course shit like the hdmi cartel would never allow a sensible technical solution to happen.
 
Back
Top