IBM Debuts Optical Transceiver Chipset

IBM debuted a new optical transceiver chipset today that researchers within the company promise will allow users to download data eight times faster than current technology. IBM cited the rising demand for digital media such as movies as the driving force behind the new technology. "IBM says it can meet that need, building its new chipset by making an optical transceiver with standard CMOS (complementary metal oxide semiconductor) technology, and combining that with optical components crafted from exotic materials such as indium phosphide and gallium arsenide. The resulting package is just 3.25mm by 5.25mm in size, small enough to be integrated onto a printed circuit board."

Re:

[Bearhouse]

Probably for the same reason that your current one got modded down too... DARPA does not finance things for the fun of it - they get first refusal on any technologies developed. With the well-documented move of big corporations, and even universites, away from 'primary' research, it's good that the state (in whatever form) stil finances this sort of stuff. Most of the recent wealth in the USA was built on intellectual capital - we can't make most stuff cheaper than the Chinese. Everybody in the economy b

Re:

[HomelessInLaJolla]

DARPA does not finance things for the fun of it

That's been my point: since DARPA is financed by taxpayer dollars, what are the taxpayers getting out of it?

But, no, I understand. It's easy to lose track of the people putting in the long hard hours at work to pay the taxes just so someone else can make three times as much working on the tax funded project. The path the money takes is just a little too complex for the majority of the Slashdot readership (and mods).

Maybe I was giving everyone more credit than they actually deserved. Perhaps I'll have t

Re:

[heinousjay]

The taxpayers are getting a general advance of technology, with attendant benefits that ripple to places no one expects. I suppose this concept goes straight past your head. I'll just point you to the Internet, which seems to be your favorite tool for disseminating half-baked rants, and let you in on the secret that its development followed the same path.

Re:

[HomelessInLaJolla]

The taxpayers are getting a general advance of technology

In what way?

I suppose this concept goes straight past your head

You're making it out to be way more than it is. The taxpayers paid for it. How do you act like IBM is doing them a favor?

disseminating half-baked rants

You can give that up now. Even MH42 [slashdot.org] has begun to realize that my observations have been correct.

let you in on the secret that its development followed the same path.

And it was taxpayer funded to begin with, and nobody is cutting the taxpayers any breaks on subscriber fees.

Re:

[heinousjay]

In what way?

Faster interconnects between components would be the obvious answer.

You're making it out to be way more than it is. The taxpayers paid for it. How do you act like IBM is doing them a favor?

I don't even know what this means. I don't think IBM is doing favors for anyone except IBM shareholders. I also think that's just fine and dandy.

You can give that up now. Even MH42 has begun to realize that my observations have been correct.

I don't know MH42. I don't care about MH42's opinions. I don't eve

Re:

[Anonymous Coward]

Hey, at least it's not my karma being ass-raped...

People, not money.

[FuzzyDaddy]

Does that mean that the taxpayers will receive a discount on the final product when it hits store shelves?

Actually, it was the citizens, not the taxpayers, that funded this. Money is collected by the government as representatives of the citizenry. Your right to control the government comes not from the money you pay into it; it comes from being a citizen.

I pay a solid middle class share of my taxes - it shouldn't give me any more say than the guy who makes nothing, or any less than the guy who pays te

Re:

[HomelessInLaJolla]

As history plays it out, though, the big businesses are more than happy to take the money from the government to fund their business and, while the executives/VPs/head researchers walk away with six and seven figure yearly salaries, the front line researchers get burned out for hardly average salaries and the technology, rather than being available freely to the citizens and taxpayers who ultimately funded it, is sold for a handsome profit first to the military (which is more taxpayer money) and then to the

Photograph..

[Arceliar]

If you want to see what it looks like, it was already featured here [dvorak.org]. The thing's damn small...

Perfect timing

[hcmtnbiker]

...and the Survey that 29% of US households dont see a need for an internet connection couldnt have been better timed. Anyone else find this slightly ironic?

Re:

[aztracker1]

WTF does a financial gap have to do with 29% of households not seeing a need for an internet connection?

I know a few wealthy doctors that can't live without their internet connection. Likewise, I know some pretty impoverished people that have the internet connection higher on their priority list than many, many other things (Hell, back in the day, some bbs sysops were pretty dirt poor)...

I think age is a much larger determining factor. The fact that many of the US's citizens are in the Baby Boomer a

Re:

[RicktheBrick]

There are several ways to entice more people to want a always on internet connection at that speed. First I see a time when hardware will be free. One will just pay for the service. There will be no local storage of data as everything will be stored at the ISP. This will greatly reduce the amount of needed storage since one copy of a movie or one copy of a software program will service thousands of people. This will end the need of either blu-ray or hd-dv since they will be stored again at the ISP on a

Wow

[CasperIV]

Your very optimistic. Of course, who ever posts that first copy of a movie that everyone references will be hit with a billion dollar lawsuit by the MPAA. They would scream bloody murder if it got even easier for people to collaborate. The biggest opponent to that Utopian ideology is going to be businesses and the conspiracy theorists.

Re:

[networkBoy]

"The biggest opponent to that Utopian ideology is going to be businesses and the conspiracy theorists."

Sadly,
I work for a large multinational in the computer industry.
I also am a privacy nut, thus you will find none of my content (unless encrypted on my local machine prior to upload) on the ISP's servers.
-nB

Re:

[osbjmg]

Not everything is about home internet usage. There is much more to networking and the internet in general, than the home user.

heheheh

[grub]

[...] will allow users to download data eight times faster than current technology.

The MPAA was not available for comment.

Re:

[Loadmaster]

The RIAA, on the other hand, had this to say:

We are ecstatic about this new development. We are currently retrofitting our squadron of lawyers with this chip so they can file lawsuits against grandmothers, children and all veterans with no legs eight times faster. Truly this is a fantastic day for us. Pretty soon we will generate more court transcripts than music.

Swi

download faster or slashdot faster?

[speculatrix]

will allow users to download data eight times faster than current

using the awesome power of slashdot it'll be possible to bring down servers at eight times the speed!

On a slightly serious note.. try asking your ISP what their contention ratio is, and their actual bandwidth at their peering points. chances are they won't tell you much detail. In practise they depend on their subscribers not trying to all max out their lines at once which is why P2P is hated by ISPs. Except for the really big companies, many organisations are probably not hosted or colocated with more than 10Mb/s or 100Mb/s anyway due to cost.

Re:

[Randseed]

The advantage, however, is that since this technology can be used to upgrade the backbones, it wlll make the ISP networks eight times faster. If the ISPs translated that to the customers at even a ratio of 6:1, then everybody would be happy: The users get 6 times the bandwidth, the ISPs have an additional 2 times the bandwidth to buffer this mess they've created by myopia.

Re:

[networkBoy]

Ironically I am hosted with three 100Mbps peering points on my little gripe site. I realize that this is a bit misleading as there are likely 3 or 4 dozen other sites on the same server sharing the same three links...
-nB

Re:

[Prune]

Uh, if chances are they won't tell me, why should I ask them? LOL

Replace your SATA cable

[Weaselmancer]

IBM tech promises 160Gb/s downloads

Net speed is nice, but I think these would also make excellent replacements for SATA. Especially when we get those nifty zero-seek time solid state flash drives. Currently, a SATA cable tops out at 3GB/s.

Re:

[Grishnakh]

So what? The disc can't physically read more than that data rate off the platters, so what's the point of having that much speed? Without some major advance, it's unlikely that hard drives will need anything faster than 3 Gb/s for a while. Maybe when we have some kind of super-fast holographic storage, this might be more important.

There's several good reasons we have SATA: it's fast enough (actually, much more than fast enough; do any drives read faster than 100 MB/s? 3 Gb/s = 375 MB/s), it's easy to us

Re:

[witte]

This is interesting tech nonetheless, i'm sure it will come in handy in a few years if produced in sufficiently high numbers to keep production cost low enough, ibm makes the comm standard open for all OEM manufacturers, and sticks to simply selling the chip in bulk.

<OTOH, this is probably just something they released from their great-tech-on-hold freezer to manipulate stock price. /paranoid>

Re:

[Grishnakh]

This is interesting tech nonetheless, i'm sure it will come in handy in a few years if produced in sufficiently high numbers to keep production cost low enough, ibm makes the comm standard open for all OEM manufacturers, and sticks to simply selling the chip in bulk.

I agree; this is why I mentioned in another post of mine that this technology makes sense for other applications, such as high-speed interconnects in a supercomputing cluster or data center. But it doesn't make any sense at all for some other t

Re:

[Spy Hunter]

Without some major advance, it's unlikely that hard drives will need anything faster

You mean a major advance like solid-state hard drives? Which, by the time this tech comes out, will have likely grabbed a significant portion of the high-end laptop market, if not more?

Re:

[Grishnakh]

Um, flash memory is pretty fast, but it's not THAT fast. It's nowhere near as fast as SDRAM, last time I checked. So I still don't see any need for a 160 Gbps optical link for storage devices anytime soon.

Optical traces

[G4from128k]

Next, we need some nifty means of printing/etching/plating optical traces on PCBs. An OGA (Optical Grid Array) would interface chips to board which would route the light to other chips.

Gallium arsenide "exotic?"

[Manchot]

Hardly. Practically every device that communicates wirelessly at microwave frequencies has GaAs amplifiers. This includes most cell phones and wireless cards.

Re:

[amchugh]

At approximately 1,000,000 metric tons production per year of semiconductor grade silicon vs. approximately 60 metric tons per year production of Gallium Arsenide, I think GaA could be considered exotic.

Re:Gallium arsenide "exotic?"

[thpr]

Practically every device that communicates wirelessly at microwave frequencies has GaAs amplifiers

Five years ago, you were right. Not anymore.

SiGe is killing GaAs [edn.com].

Many of the devices communicating in the higher frequences of the microwave range are based on Silicon Germanium. This includes cell phones [rfdesign.com].

Almost ALL WiFi radios are SiGe [ralinktech.com] [PDF warning]. Some have even moved to RFCMOS [wi-fiplanet.com].

Most GPS devices are SiGe [motorola.com].

Oh, and TV Tuners, too [mwee.com].

Gallium Arsenide *is* exotic, because it has to be done in specialized fabs, not those that run silicon wafers. That significantly drives up the cost vs. SiGe and RFCMOS.

Re:

[networkBoy]

Nevermind that GaAs is very fragile and AFAIK 4" wafers are still the norm, while Si fabs are pushing 12".

-nB

Re:

[WhoBeDaPlaya]

GaAs is incredibly expensive. You do know that IBM bandgap engineered SiGe to get away from that right ;)

So in other words...

[dgatwood]

From this writeup, I'm having a hard time seeing h ow this differs significantly from an LED. What am I missing?

Re:

[Workaphobia]

I don't understand how this is different from fiber optics as we know it. I was under the impression that you generally had two big-ass expensive routers miles apart, some optical cable between them, and some pricey interface hardware on either side to send and receive information at gigabits a second. From this article it sounds like IBM just invented light.

Demand for bandwidth drives invention!

[RingDev]

Anyone else think the arguments against Net Neutrality just got a little weaker?

-Rick

Re:

[fyngyrz]

Anyone else think the arguments against Net Neutrality just got a little weaker?

No. The actual arguments based on greed, not bandwidth. Technical arguments against net neutrality are simply fodder for the common person to argue about. All decisions will be taken based upon degree of profit that appears to be available.

Speed Now

[Doc Ruby]

That transceiver is good for 160Gbps, in 2010, when it might be for sale.

Meanwhile, what do you do when you need more than 10Gbps? Stuff a PCIe bus with 2x10Gbps boards? Spend a $million on an experimental 100Gbps transceiver?

It's weird that there seems to be $10 1Gig-e, $450 10Gig-e, $750 2x10Gig-e, and then... nothing. Since even PlayStations include 1Gig-e, surely the horizon isn't really just 10x that speed?

Re:

[cheater512]

You dont have a gigabit network do you? If you did you'd be demanding more speed than the pathetic 120mb/s you get with gigabit.

And thats with most computers needing a minute or two after a copying a file just to recover from the ordeal of having that much data shoved in to it at once..

Re:

[Wesley Felter]

For long-haul links there's 40Gbps OC-768 and for data centers there's 2x16Gbps Infiniband.

Ethernet likes to increase by factors of 10 but 100Gbps is not practical yet, so there's nothing in between.

Re:

[Doc Ruby]

Thank you for discrediting yourself in politics, technology and rhetoric in a single post, bitch:

PCI Express [wikipedia.org] (Official abbreviation PCIe, PCI-E often used, not to be mistaken for PCI-X) is a computer system bus [...]

I bet you voted for Bush a couple of times, too.

And you probably have some kind of inane, dishonest excuse that you're somehow still not wrong in spitting pure bullshit like you just did. Spare me.

Re:

[Brane2]

Well, Wikipedia is WRONG. PCIe is not "bus", since it is strictly point-to-point.

Re:

[Doc Ruby]

No, you are WRONG.

Because the point-to-point topology of PCIe is dynamically configurable through the PCIe switch, making its points selectable at will. It's backwards compatible with PCI, which is indisputably a bus, so new devices can treat it as a bus.

Ethernet is still a bus when it's run through a switch.

Get a grip on what these terms mean now, in practice, not what some abstract semantics say they'll mean when they're invented in a lab.

Get on the Bus

[Doc Ruby]

More of the PCI Express experts who are somehow "wrong" while you, a nobody, are somehow right. Let Jon Stokes explain it [arstechnica.com] to you:

In a point-to-point bus topology, a shared switch replaces the shared bus as the single shared resource by means of which all of the devices communicate. Unlike in a shared bus topology, where the devices must collectively arbitrate among themselves for use of the bus, each device in the system has direct and exclusive access to the switch. In other words, each device sits on its

Re:

[pr0nboy]

Okay- then why don't you point out anything in the PCIe spec that refers to it as a bus (I'll accept anything from the 1.0, 1.0a, 1.1, or 2.0 specs). They use the word "link" religiously. I wonder why that is...

This will be fun.

I think I may be missing something

[WombatDeath]

A PC using that board would be able to reduce the download time of a typical high-definition feature-length movie from 30 minutes to one second, the company said.

That's nice, but I don't know of anyone able to provide me with that movie in one second, much less anyone with the bandwidth to receive it (or write it to disk) that quickly. The bottleneck in my downloading experience sure as hell doesn't exist within my beige box.

Will this actually be useful for anything in 2010?

Re:

[brunascle]

my thoughts exactly.

until ISPs offer much more bandwidth, the only place i see this being useful is on a LAN.

Re:

[heroofhyr]

A PC using that board would be able to reduce the download time of a typical high-definition feature-length movie from 30 minutes to one second, the company said.

Jesus Christ. Only 30 minutes for a HD full-length movie? Someone find out where IBM is downloading its torrents.

That would be groovy

[WombatDeath]

"Sorry, can't stop - have to take my seven octillion Libraries of Congress for a walk".

Re:

[stud9920]

Obligatory "Windows Vista will now download in less than 2 days" post

Hah.

[rackhamh]

What four-letter word best describes what this technology will be used to access?

"data"

What?

How is this useful for the near future?

[Grishnakh]

I actually read (quickly) TFA, and I don't see how this is useful at all for the applications they're envisioning. In a high-performance data center or supercomputer cluster, sure, 160 Gbps links might be quite useful. But for connecting homes to the internet? Sorry, I don't see it.

For one thing, this technology is far faster than anything we already have, or what anyone is demanding. Add up a fast internet connection, VoIP, and a few TV channels in HD, and you still don't come close to needing 160 Gbps

Re:

[SwedishPenguin]

Obviously it wouldn't be used to connect peoples homes (at least not in the foreseeable future), but ISP's and backbone providers could probably have some need for this, especially considering that bandwidth to customers is increasing. I have a 100 Mbps fiber connection in my apartment, it would "only" take a residential area of 1600 such apartments downloading/uploading at full capacity to max out this chipset. Now imagine this on a bigger scale, hundreds of thousands of households, a medium sized city. Th

Re:

[Grishnakh]

The whole origonal article seems pretty brain dead. Its comparing current WAN speeds with new fibre LAN speeds, and, it uses as an example appliction -- dowloading a movie -- something thats definately a "home" user application.

I agree on the brain dead part. The reporter is an idiot, trying to make a technology useful in a data center relevant to people who he thinks have no concept of things other than downloading movies at home. Most people have heard of data centers by now, especially anyone reading a

Buying up the "old" ones on E-bay

[aegl]

So this new thing can download 8-times faster than the previous generation.

It will take just one second to download a complete HD movie.

I think I can survive with waiting 8 seconds to download a movie (it will take me 90+ minutes to watch it anyway).

So I'll be looking to buy up some of the "old" cards when people toss them out to upgrade to these new cards.

160Gb/s may be limited to on-campus distances

[viking80]

At 160Gb/s, each bit is about 1mm long in the fiber. The dispersion of a fiber will smear away the eye in a kilometer. Using zero dispersion fiber causes problems with DWDM, so this may only be usable in a data center.

Since bundles 10Gb/s X 16 are available as single plug, there will be little practical difference for users unless it is cheaper than the 10Gb/s X 16 bundles.

Re:

[owlstead]

"At 160Gb/s, each bit is about 1mm long in the fiber. The dispersion of a fiber will smear away the eye in a kilometer. Using zero dispersion fiber causes problems with DWDM, so this may only be usable in a data center.

Since bundles 10Gb/s X 16 are available as single plug, there will be little practical difference for users unless it is cheaper than the 10Gb/s X 16 bundles."

Well, that seems to be the idea of the article. If there is any way to bring our company network up to this speed at low cost, I would

I Don't Get It

[Nom du Keyboard]

allow users to download data eight times faster than current technology. IBM cited the rising demand for digital media such as movies as the driving force behind the new technology.

I don't see how this is going to make my cable connection run any faster, which is the only part of downloading movies faster that would have any effect on me.