• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

IBM creates cheap, commercially viable 'electro-photonic' integrated chip

Status
Not open for further replies.

Kinitari

Black Canada Mafia
I tend to think of myself as a comp-sci guy, and while I have a general idea of what this means and why it's awesome, it's still pretty hard for me to wrap my head around. Anyone wanna take a stab at explaining it? Because I am not confident enough to do so.

http://www.extremetech.com/computin...commercially-viable-silicon-nanophotonic-chip

IBM has become the first company to integrate electrical and optical components on the same chip, using a standard 90nm semiconductor process. These integrated, monolithic chips will allow for cheap chip-to-chip and computer-to-computer interconnects that are thousands of times faster than current state-of-the-art copper and optical networks. Where current interconnects are generally measured in gigabits per second, IBM’s new chip is already capable of shuttling data around at terabits per second, and should scale to peta- and exabit speeds.

...

The payoff makes all the hard work worthwhile, though. IBM now has a cheap chip that can provide a truly mammoth speed boost to computers. It’s not too hyperbolic to say that this advancement will single-handedly allow for the continuation of Moore’s law for the foreseeable future.

In these chips, there are optical modulators and germanium photodetectors that can send and receive data at 25 gigabits-per-second (Gbps), using four-channel wave-division multiplexing (WDM).

...

While we couldn’t even get a ballpark figure out of IBM, the use of a standard 90nm process means that these chips probably cost no more than a few dollars to produce. IBM is targeting super and cloud computing first, where bandwidth between nodes is a serious bottleneck — but there’s no reason that these chips won’t eventually find their way into consumer hands.

...

This is the chip that could power the next-generation optical interconnect between your desktop’s CPU, GPU, and RAM. This is the chip that could directly wire your PC into your ISP’s fiber-optic network, potentially unleashing terabit-or-higher download speeds. This chip is a big deal.

Of course, while the article says that these 'could be in the market within the next couple of years' - there is no guarantee. But IBM is looking to target super computers first, which will be pretty interesting considering that they have some pretty baller super computers of their own.
 

Tacitus_

Member
The big thing about this news is that they managed to convert their manufacturing process to churn out these chips that have both chip types in the same chip. It's officially out of the labs and could start up production reasonably soon.

E: Like so:
Second, and perhaps more importantly, IBM has manufactured these chips on its 90nm SOI process — the same process that was used to produce the original Xbox 360, PS3, and Wii CPUs. According to Solomon Assefa, a nanophotonics scientist at IBM Research who worked on this breakthrough, this was a very difficult step. It’s one thing to produce a nanophotonic device in a standalone laboratory environment — but another thing entirely to finagle an existing, commercial 90nm process into creating something it was never designed to do. It sounds like IBM spent most of the last two years trying to get it to work.
 

GraveRobberX

Platinum Trophy: Learned to Shit While Upright Again.
The series of tube that are the internet are getting streamlined

Tubes will not get clogged, EFFICIENCY!
 
KyV08.png
 

Nesotenso

Member
metal interconnects are used between on-chip elements which transfer electrons/ charges between them. IBM has successfully replicated the 90 nm manufacturing process for optical interconnects which transfer photons
 

nitewulf

Member
basically if you can think of the bus speeds between the various chipsets in your computer as the data pathways, they are made of copper and bandwidth limited, or even within the same chipset if one side has to talk to the other side, the communication is via copper links. IBM devised a way to replace the copper links with fiber optic links (I havent yet to read the artcile, but the idea is not new). So the data/communication bandwidth between the chips and different parts of the computer is basically raised to another level.
 

Zaptruder

Banned
So which bottlenecks do these things obliterate in computer architecture?

The way I'm reading it, almost makes it sound like they can replace copper with this? That doesn't make sense.

It'd be cool if all computer components could essentially act like one giant chip because the connections were that fast... but I don't think that's what this means.
 

tsef

Member
What's the benefit we'll see in terms of DBZ power levels?

Probably over 9000.

Though seriously it could be a big deal. But I'm always afraid of the potential for this to be a vapourware. So many science article read more like hype than science nowadays.
 

TCRS

Banned
Probably over 9000.

Though seriously it could be a big deal. But I'm always afraid of the potential for this to be a vapourware. So many science article read more like hype than science nowadays.

After reading a couple of articles on German websites I was about to write that. None of them sound half as enthusiastic as the article above.
 

leroidys

Member
WOWOWOW. Laygaf, bus speed is the biggest bottleneck on most computers, (well this + electromechanical disks) this is fn huge.

The payoff makes all the hard work worthwhile, though. IBM now has a cheap chip that can provide a truly mammoth speed boost to computers. It’s not too hyperbolic to say that this advancement will single-handedly allow for the continuation of Moore’s law for the foreseeable future.

This is the first thought I had.
 

1-D_FTW

Member
Probably over 9000.

Though seriously it could be a big deal. But I'm always afraid of the potential for this to be a vapourware. So many science article read more like hype than science nowadays.

Every time I read a Wired article, I think this same thing. It's like the National Enquirer branched off into science.
 

Kinitari

Black Canada Mafia
Probably over 9000.

Though seriously it could be a big deal. But I'm always afraid of the potential for this to be a vapourware. So many science article read more like hype than science nowadays.

After reading a couple of articles on German websites I was about to write that. None of them sound half as enthusiastic as the article above.

Every time I read a Wired article, I think this same thing. It's like the National Enquirer branched off into science.

I think it's hard to sort through the good stuff and find the bad stuff - but after enough time you can sort of see which things are the things to really be excited about. I think this is something that is 'exciting' - but might sound boring. I know that bus speed bottlenecks are a big deal, so if this can pretty much eliminate them, it should have a cascading effect.

This definitely isn't vaporware either - vaporware in the tech world isn't as common as people think - a lot of this stuff just takes a lot more time than expected.
 

Mudkips

Banned
Like so:

Current copper interconnects
Cheap and easy, not very fast
zwBJg.jpg


Current optical interconnects
Expensive and you have to go to the airport first, but very fast once you're in the air
QG2jM.jpg


New hotness
S1tYU.jpg
 

Kinitari

Black Canada Mafia
This, what real world implications does this have?

Well, all my limited mind can only really think of is that it would REALLY help for eventually upping the download cap - the process of converting a optical signal to something you can actually interact with creates a bottleneck - this would essentially remove the bottleneck. So, the processing of large amounts of data, regardless of location (because it could move through a fibreoptic connection) wouldn't be as difficult. For a more concrete example of why this is good, I can imagine this playing a big role in cloud computing - and even cloud processing (maybe a bunch of super computers can team up on large problems with this!).

I'm sure there are more, and I am sure I am sorely misrepresenting the potential - but it's above my head!
 
From the sounds of things, this tech can do the following.

1: Speed up bus speed by several multitudes
2: By changing materials you produce a cooler chip allowing for higher overall clock speeds
3: Allow multiple signals through the same chip lines
4: Can integrate network communication aspects directly into the chip as it can communicate via optical like networks

In a nutshell, a far more efficient chip that whats out there so far

(I'm no engineer or technology guru, thats just my guess from what I'm reading up so feel free to correct me if i'm wrong)
 

The Technomancer

card-carrying scientician
Incredible. I have a lot of respect for IBM. Like 3M they're an enormous corporate entity that still works on really fundamentally innovative stuff.
 

leroidys

Member
Faster computers? Does anyone ever really know what's going to be done with faster computer technology until the programmers get their hands on it and actually make stuff for it?

It's not changing the architecture of the computer, it is just changing how fast parts of your computer can talk to each other. The exact same algorithms would run MUCH faster on a computer with these kinds of connections.
 

SRG01

Member
I don't think interface bandwidth is a serious bottleneck in consumer applications. This just means it will not be a bottleneck for some time to come.

Actually it is. Bus widths and speeds will always be limiting factors inside and between chips.

Not to mention that this might actually reduce thermal requirements of various devices too.
 

Zaptruder

Banned
Wait... so you guys are saying that these optical interconnects can be built on the PCB layer?

Because isn't that stuff all part of the bridge that connects chips with the physical sockets (like PCIE slots and CPU sockets)?

That's pretty crazy nuts. It's almost like enabling a different form of computer architecture and design entirely if that's true.

It's almost like the contents of an entire motherboard becomes one giant intergrated chip.

L2, L3 cache on CPU? Why not have 16+GB of ram act at cache like speeds?
 

DonasaurusRex

Online Ho Champ
Hmmm, will this mean I can play DOTA2 on USW from australia with a lower ping? :U

lol no australian ISP's suck dick bro , foolish dreams lol....

ok ok maybe i mean if they spend money upgrading to these chips because they save money to run their business...maybe your new ISP will give you another teaspoon of bandwidth. Otherwise your network is going to be what your ISP allows you to have man no matter what your comp / network can handle.
 
Status
Not open for further replies.
Top Bottom