By Luke Lattanzi | Staff Writer
In collaboration with researchers from Oregon State University, Dr. Alan Wang — the Mearse chair in biological and biomedical engineering — had his latest breakthrough in computer chip energy listed among the “Top 5 Research Stories of 2023” by the National Science Foundation.
The joint research effort led to a breakthrough in developing a prototype photonic computer chip that can better regulate its temperature, leading to more efficient energy usage.
Photonic computer chips are commonly used in data centers and supercomputers around the country. They are different from typical electric computer chips in that they use optical fibers, which use photons — or light particles — to convey information.
Data centers, such as ones operated by technology companies like Google, account for roughly 2% of all energy used in the U.S., according to the National Science Foundation. However, according to Wang, that number is actually higher when accounting for the overall energy used to maintain those facilities.
Wang said the problem with current photonic computer chips is that they must be cooled by external methods, such as air conditioning, in order to continue functioning, leading to more energy consumption overall.
“A data center itself may consume 2%, but if we are considering overall energy … when a computer is running, it generates a lot of heat, so you need to cool down the entire facility, and that uses a lot of energy too,” Wang said. “Most of the energy, when you do computation, the energy is dissipated into heat. A simple example: If you wanted to send information through a copper wire, the current flows through the copper wire, and the copper has a resistance. A big portion of [that] energy is turned into heat, and that increases the temperature.”
While all photonic computer chips currently use optical fibers to process information, the prototype Wang’s team developed is able to convert electrical signals from computers into photonic signals, significantly reducing the amount of energy needed to operate them. The optical fibers are very transparent, allowing the signal to travel at high speeds with minimal energy consumption.
“For data centers and supercomputers, the bottleneck is actually the link between different computing units,” Wang said. “For example, if you want to send information from one computer to another, you cannot use an electrical cable, because the bandwidth is too low and the energy consumption is too high. So if you want to send high-speed information among different computation units, you have to use optical fibers.”
However, even with this breakthrough, Wang said there are several challenges when it comes to taking a prototype and turning it into a product that can be used in real-world situations.
“In order to be used in the product lines, there are many things you need to satisfy, like reliability, compatibility, cost,” Wang said. “It’s very difficult to meet all the metrics. Sometimes you focus on the particular problem and you can solve it, but the other constraints — like the cost, the compatibility with the system and the long-term reliability — may limit the application of the technology. It’s the same for me too. We demonstrated a prototype, but there is still a long way to go to turn it into a real product.”
Dr. John Conley, a professor of electrical engineering and computer science at Oregon State University and a co-author of the research, said the prototype is especially important due to how quickly energy demands for computing technologies are rising.
“The amount of energy the world is using for computing technologies has been increasing exponentially,” Conley said. “It’s actually increasing at a very steep exponential rate toward the overall worldwide energy consumption, which has been increasing slowly. But within a few years, it will catch up, and basically, our computing technology can’t use all the energy in the world for Facebook. So eventually that curve will reach up to the worldwide energy production and basically be limited by that.”
Despite the logistical challenges of taking a prototype and turning it into a product that can be used in real-world applications, Conley said the research is nevertheless very promising, and he is optimistic about what will become of it.
“You never know what’s going to catch the imagination [of the National Science Foundation] or the popular press,” Conley said. “But the fact that the [National Science Foundation] picked it up as one of their top stories alongside black holes and plastic eating bacteria — that’s pretty awesome.”