facebook rss twitter

IDF Spring 04: R&D Briefing - Silicon Photonics

by David Ross on 16 February 2004, 00:00

Tags: Intel (NASDAQ:INTC)

Quick Link: HEXUS.net/qawd

Add to My Vault: x

IDF..... Press Pre-Brief.

IDF does not offically start until Tommorow however Intel today decided to spend time with the press going over current RandD projects. The first is that of Silicon Photonics, the talk was very similar to that of last time however it is getting more interesting as they shed more light on the situation (no pun intended.).

So on with the optical converts....
The Internet is a huge communication network – fibre, copper and other methods, that is before it even hits your home or office – then it hits more copper or wireless networks. However the barriers are slowly needing to be broken, and pipes are going to one day hit capacity.

Last year at IDF Intel discussed in great detail about a new research department which had been established within the communications lab – Silicon Photonics.

Photonics is the technology of emission, transmission, control and detection of light (photons) – in the form of fibre optics and opto-electronic communication.

However today’s major draw backs is the fact that the photonic devices are typically not made in or with silicon – thus the cost structures are high and the integration is harder, some of these are made individually and this brings huge cost implications.

Silicon Photonics is the research efforts to develop photonic devices using silicon as a base material and do this using standard, high volume silicon manufacturing techniques.

Today Intel have released a scalable 1GHz silicon modulator. Previous modulators were held back to the 20Mhz frequency. The microprocessor performance in turn has got faster and faster thus it demands higher speed interconnects and more bandwidth.

Following this chain of thought the communications which are needed also need to scale accordingly. I remember when I used my first Gigabit configuration and the speed was blinding – in 10 years time this will, of course, be regarded as slow.

Together with the increase in transistor counts in relation to Moores law, Communications have followed this trend but they have hit the ceiling of 10Gbit/s currently. The aim is to bring optical to the masses, in an integrated package. Optical technology can be used on data buses – directly delivering and removing the data from the silicon. The aim is to reduce the gap between photons and electrons.

Previously you would require a modulator between the previous network configurations, and this has had to change – Intel had to find a way to carry the data from the silicon [Processors and core logic]. This change has come around with a transistor-like phase shift device which does direct coding in to the light beam. This signal is then decoded at the other end. Both the encoder and decoder is integrated in to the silicon package.

The plan is to transfer as much photonic ‘fibre’ communication to silicon as this will drive down costs and enabled mass market deployment due to the cost cutting. ”The aim is to open up new markets for current markets, with the convergence with price points to enable more people to use it.” said Pat Gelsigner – Intel CTO.

This enables Intel to drive down the costs via scales of economy (manufacturing in a fab is far cheaper than making current non silicon units.)

Within the presentation which we saw today, Intel demonstrated the first 1GB/s data modulator which is running on a normal PCB. They used this to move high definition film across the digital modulator is something which wasn’t possible with the previous technology – the delivering of high definition video will be significant within the video and home streaming convergence.

To be able to do siliconize the optical you need to:

Have a light source via a low cost laser, the guide light is provided via silicon wave guides, the modulation is done via a silicon modulator, the light detection is via a silicon photo-detector, lowcost assembly would be done via Passive alignment, and finally intelligence which is where the CMOS circuitry comes in to play. All of these are the essential building blocks.



The secondary advantage of optical communication is the reduced EMI and cable weight, but the primary benefits are of higher speeds over longer distances. For example, on a laptop there is currently a distinctive ribbon cable between the monitor and the base of the laptop, this is the way that the graphics card sends information to the screen. However if the cost structure came in to play then this could be reduced and we could see the implementation of antennas within the screen for WiFi and Bluetooth. I wonder how long until we see 1GB WiFi communication?

The future is going to be driven on bandwidth and the processing of the data. Simply, silicon photonics will help to put communication in place.

This might all sound far too technical and indepth but this technology will have an impact on the future work which you will do with your system. Not now, but maybe in 10 years – Intel currently have no technology deployment roadmap, but they do have functioning technology within the labs.

Is this the start of the end of silicon? This is not the case, according to Intel. They believe it still has some headroom – no need to drive biological and other substrates. Even with the optics and similar technologies Intel believes that they will continue to use silicon and CMOS manufacturing technology.



HEXUS Forums :: 0 Comments

Login with Forum Account

Don't have an account? Register today!
Log in to be the first to comment!