Intel solves the silicon laser issue by going hybrid--adding InP lasers to its silicon chips. That sounds exotic and expensive, but Intel claims to be perfecting a wafer-scale technology for inexpensively adding InP lasers anywhere you want on a chip:
I agree this is an exiting research. But I also agree with Rich: this is very far from being practical. All these announcements about silicon photonics neglect one fundamental problem: you can't build a laser in silicon. So you can switch, guide, detect, etc. but how are you going to generate the signal??? Anyone? Kris
Silicon photonics is real--in the lab--and will happen eventually, but the III-V semiconductor makers are not going down without a fight. In some ways its a disadvantage to hawk the lower cost of silicon photonics, because the reason designers choose gallium arsenide is often because they are willing to pay for its better performance. These silicon photonics innovations will catch on, but they have to get the performance up before designers will use them.
Whenever I see news about silicon photonics, I get excited because I'm at a CMOS silicon IC design company (Broadcom), and while we have some business having to do with fiber optics, it's on the periphery, to say the least. I'm looking forward to the day when I can put my fiber background together with the electronics. I have to ask whether the rubidium gas is compatible with CMOS processing, though. Man, I can't wait for this stuff to happen for real!
Blog Doing Math in FPGAs Tom Burke 15 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...