Sun Working on Replacing Wires with Lasers to Drastically Increase Chip Speeds

Illustration for article titled Sun Working on Replacing Wires with Lasers to Drastically Increase Chip Speeds

If our computers are ever going to hit speeds that'll allow us to do things like simulate the big bang, create artificial intelligence or create giant, building-sized robots to fight for our amusement, we need to move beyond wires. After all, their physical makeup is going to hit a wall at some point, so rather than just continuing to work at making faster wires, we need to look at what's going to come after wires, and that something is lasers.


Sun Microsystems has just received a $44 million contract from the Pentagon to do just that. They're to work on a way of connecting silicon chips via lasers, which, if successful, will increase chips speeds by a factor of thousands.

Computer scientists have long sought a way to make faster and cheaper computers by making larger chips on a single wafer of silicon, a manufacturing process called "wafer scale integration." If the Sun researchers' idea can be proved technically feasible and manufactured commercially, it would be possible to create more-compact machines that are a thousand times faster than today's computers, the company said. Each chip would be able to communicate directly with every other chip in the array via a beam of laser light that could carry tens billions of bits of data a second.


The only problem is that Sun says that they are only expecting a 50% success rate, so this advancement could be a lot time coming. But hey, they're working on it, so those giant robot fights might be closer than we could have ever imagined. [NY Times]

Share This Story

Get our newsletter


Lasers are not efficient for very short-haul communications. To communicate from chip-to-chip you need to turn high-speed electrical signals in to light and then back again. This is inefficient and generates heat.

If you can communicate from the silicon to the laser diode with electrical signals, why not go a little further and send the signal all the way to a neighboring chip?

While high-speed electrical signally has its challenges, there are techniques that could take it further than today's limits (of around 10Gbps for a SERDES link).

I would give them less than a 50% chance of success!

Incidentally, this is the same team that worked on capacitative chip-to-chip communications. While interesting, that had a number of challenges and to the best of my knowledge has yet to find its way in to any real products.