Institute Breaks Transmission World Record in an Optical Fiber with 125,000 Gbps

[BiG StroOnZ]

 

Researchers from the Network System Research Institute of the National Institute of Information and Communications Technology (NICT, Japan) have achieved the world's first transmission exceeding 1 petabit per second (125,000 Gbps) in a single-core multi-mode optical fiber.

 

Quote

 

 

"To date, transmission experiments in optical fibers supporting large number of modes were limited to small optical bandwidths. In this study, we demonstrated the possibility of combining highly spectral efficient wideband optical transmission with an optical fiber guiding 15 fiber modes that had a cladding diameter in agreement with the current industry standard of 0.125 mm," wrote the researchers. To conduct their experiment, the researchers developed a wideband transceiver subsystem to transmit and receive several hundred highly spectral efficient WDM (Wavelength Division Multiplexing) channels of high signal quality. It was this transmission system that made use of a multi-mode fiber that demonstrated the record-breaking transmission speed. The success of this experiment is expected to advance high-capacity multimode transmission technology for future high capacity optical transmission systems. Now, the researchers have plans to extend the distance of large-capacity multi-mode transmission and integrate it with multi-core technology to establish the foundation of optical transmission technology with increased capacity.

 

Source 1: https://www.nict.go.jp/en/press/2020/12/18-1.html

Source 2: https://interestingengineering.com/institute-breaks-transmission-world-record-with-125000-gbps-using-an-optical-fiber

 

This is an extremely incredible achievement, as 125,000,000 Mbps is an astronomical number. Reminds me of the days of early broadband when I used to dream of T1, T2, and T3 connections. But this is on a whole other level of dreaming, seriously. From what I can see, this supposedly surpasses the current record transmission in a multi-mode fiber by nearly 2.5 times! For ease of understanding: this record converts to 15,625 gigabytes per second (or 15.625 terabytes per second)! 

 

Side Note:

 

23 hours ago, wanderingfool2 said:

haha, I originally thought the same.  @BiG StroOnZ didn't mention it, but the 125tbps rate was per mode, and they used 15 modes through the multi-mode fiber (which apparently wasn't as achievable in the older technologies that achieved speed).  So it gets to petabits by multiplying it by 15

 

18 hours ago, wanderingfool2 said:

Hoping that @BiG StroOnZ edit's his original to be a bit more clear of the 125,000 gb/s (*15 modes to equal a petabit).  1petabit/s = 125TB/s 

 

[Doobeedoo]

Amazing and everything. Just sucks how most will still be stuck on old lines for who knows how long in general.

[Lurick]

An awesome achievement indeed, now if only I could wake up and realize I converted things wrong in my head for a moment, lol.

[Lurick]

7 minutes ago, Doobeedoo said:

Amazing and everything. Just sucks how most will still be stuck on old lines for who knows how long in general.

Because $$$

I'm sure this will make its way into the enterprise space in the next 5-10 years so maybe in 100+ years the end consumer will benefit, lol. (and Comcast will still limit you to 1TB of data)

[starry]

Why dont they just make a zip file and attach it to an email?

[Doobeedoo]

4 minutes ago, Lurick said:

Because $$$

I'm sure this will make its way into the enterprise space in the next 5-10 years so maybe in 100+ years the end consumer will benefit, lol. (and Comcast will still limit you to 1TB of data)

I mean, I'd just want a basic fiber optic like and speeds though. We know ISPs slack and are greedy. I'm not in US though, no limits.

[trag1c]

3 hours ago, huilun02 said:

About as significant as someone getting a CPU overclock world record on LN2

 

Having mini nuclear powerplants to power all our vehicles and electronic devices would revolutionise the world. Its frankly not that far off and I can't wait to be fiddling with mini nuclear fuel rods.

So what you're saying is that when I am working on my own vehicles I should probably leave the BFH (big fucking hammer) in tool box and try to persuade things another way.

 

 

As far OP

 

It's cool but trying to achieve that speed at any sort of usable distance is going to be damn near impossible.

[WereCatf]

6 hours ago, BiG StroOnZ said:

exceeding 1 petabit per second (125,000 Gbps)

Shouldn't that be 1 terabits per second?

[wanderingfool2]

18 minutes ago, WereCatf said:

Shouldn't that be 1 terabits per second?

haha, I originally thought the same.  @BiG StroOnZ didn't mention it, but the 125tbps rate was per mode, and they used 15 modes through the multi-mode fiber (which apparently wasn't as achievable in the older technologies that achieved speed).  So it gets to petabits by multiplying it by 15

[WereCatf]

Just now, wanderingfool2 said:

haha, I originally thought the same.  @BiG StroOnZ didn't mention it, but the 125tbps rate was per mode, and they used 15 modes through the multi-mode fiber (which apparently wasn't as achievable in the older technologies that achieved speed).  So it gets to petabits by multiplying it by 15

Well, that explains it, but then it's just stupidly worded. I'm not sure that's any better.

 

Note, I am not blaming @BiG StroOnZ for this.

[straight_stewie]

4 hours ago, trag1c said:

It's cool but trying to achieve that speed at any sort of usable distance is going to be damn near impossible.

TL;DR if the latency can be reduced to a useful point, this could open up a lot of possibilities in local networks for large installations.

 

23 kilometers (14 miles) is pretty good to start with. I mean, it's short, but still useful in some areas with closely packed nodes (like bigger cities) without any improvement in transmission distance. Yeah, that has limited usefulness, but higher bandwidth would also reduce latency in high load scenarios (like bigger cities).

More in the present, it by far beats out infiniband bandwidths. If the latency is low enough, this would probably start being used in new supercomputer/datacenter/cloud-center installations pretty quickly. If the latency can be reduced to the single digit nanosecond scale for local networks (like what you would find in one of those installations), that could make networked processor-to-processor connections possible.

 

For reference, 1 petabit per second is 125,000 GB per second. As a reference frame, read bandwidths of L3 caches on modern processors are around 370 GB per second, so this is roughly 378 times faster than L3 cache read bandwidth. Of course, that's not quite an apple-to-apples comparison, but I suspect that it still beats out read bandwidth of modern L3 caches when you account for the overhead of building the data packets (especially if one can assume that this is a feature people want, so will be integrated with on die hardware).

[wanderingfool2]

24 minutes ago, straight_stewie said:

For reference, 1 petabit per second is 125,000 GB per second. As a reference frame, read bandwidths of L3 caches on modern processors are around 370 GB per second, so this is roughly 378 times faster than L3 cache read bandwidth. Of course, that's not quite an apple-to-apples comparison, but I suspect that it still beats out read bandwidth of modern L3 caches when you account for the overhead of building the data packets (especially if one can assume that this is a feature people want, so will be integrated with on die hardware).

Hoping that @BiG StroOnZ edit's his original to be a bit more clear of the 125,000 gb/s (*15 modes to equal a petabit).  1petabit/s = 125TB/s ;)

 

I could see this being more practical on the undersea cables first, as those ones cost an arm and a leg to add capacity via wires...then again, this is just all in the lab kind of thing...but given the amount of undersea cables, maybe it doesn't have to be commercially viable.

 

[straight_stewie]

1 hour ago, wanderingfool2 said:

I could see this being more practical on the undersea cables first, as those ones cost an arm and a leg to add capacity via wires...then again, this is just all in the lab kind of thing...but given the amount of undersea cables, maybe it doesn't have to be commercially viable.

At only a 23km (14 mile) transmission distance, and with transmission distance being inversely proportional to bandwidth in fiber optics, I feel that it will be a very long time before this becomes viable for trans-oceanic cabling, where you need 100s of miles of cables between repeaters.

[leadeater]

13 hours ago, starry said:

Why dont they just make a zip file and attach it to an email?

Spanned floppy disk archive would be faster, USPS will get it there in no time!

[wanderingfool2]

9 hours ago, straight_stewie said:

At only a 23km (14 mile) transmission distance, and with transmission distance being inversely proportional to bandwidth in fiber optics, I feel that it will be a very long time before this becomes viable for trans-oceanic cabling, where you need 100s of miles of cables between repeaters.

Well from what I gather, the 23km is just the testing...I am guessing it would be capable of longer distances.  (I'm guessing they probably are limited to how many km of cable they have access to).

 

By the little research I did on this (as in haven't really done much to vet what I am about to say).  It seems like the longest cables from 2014 are 380km, but still had unpowered repeaters every 100km.

 

Either way, this is just technology in the lab still