{"id":69475,"date":"2018-04-18T11:00:05","date_gmt":"2018-04-18T15:00:05","guid":{"rendered":"http:\/\/www.bu.edu\/eng\/?p=69475"},"modified":"2024-03-07T11:22:10","modified_gmt":"2024-03-07T16:22:10","slug":"photonic-communication-comes-to-computer-chips","status":"publish","type":"post","link":"https:\/\/www.bu.edu\/eng\/2018\/04\/18\/photonic-communication-comes-to-computer-chips\/","title":{"rendered":"Photonic Communication Comes to Computer Chips"},"content":{"rendered":"

<\/em>Startup\u2019s optoelectronic chips could reduce energy usage by up to 50 percent in data centers while increasing computing speeds.<\/strong><\/p>\n

By Rob Matheson – MIT News Office, originally published on\u00a0MIT News<\/a>.<\/em><\/p>\n

\n
ECE Professor Milos Popovic moved to BU in mid-2016 from a previous position as Assistant Professor at University of Colorado Boulder. While at CU Boulder, Popovic\u2019s research group teamed up with research groups at MIT and UC Berkeley to work on the DARPA-funded Photonically Optimized Embedded Microprocessors (POEM) research program, work that continued after his group’s move to BU and completed in November 2017. The collaboration lead to the launch of Ayar Labs, Inc., where Popovic is one of the six co-founders. Since joining Boston University, Prof. Popovic has leveraged his affiliation to benefit BU students by opening doors to Ayar Labs’ internship program. Ayar Labs President, Chief Scientist and Co-Founder Dr. Mark Wade is one of Popovic’s first PhD graduates.<\/span><\/h6>\n<\/div>\n
<\/div>\n
\n
\"MIT-Ayar-Labs-01_0\"
Ayar Labs’ optoelectronic chips move data around with light but compute electronically.<\/figcaption><\/figure>\n<\/div>\n

 <\/p>\n

ECE Professor Milos Popovic moved to BU\u00a0in mid-2016\u00a0from a previous position as Assistant Professor at University of Colorado Boulder. \u00a0While at CU\u00a0Boulder, Popovic\u2019s research group teamed up with research groups at\u00a0MIT\u00a0and UC Berkeley to work on the DARPA-funded Photonically Optimized\u00a0Embedded Microprocessors (POEM) research program, work that continued after his group’s move to BU and completed in November 2017. \u00a0The\u00a0collaboration lead to the launch of Ayar Labs, Inc.,\u00a0where\u00a0Popovic is one of\u00a0the\u00a0six co-founders. \u00a0Since joining Boston University, Prof. Popovic has\u00a0leveraged his affiliation to benefit BU students by opening doors to Ayar Labs’ internship program. \u00a0Ayar Labs\u00a0President, Chief Scientist\u00a0and Co-Founder Dr. Mark Wade is one of Popovic’s first PhD graduates.<\/p>\n

With novel optoelectronic chips and a new partnership with a top silicon-chip manufacturer, MIT spinout Ayar Labs aims to increase speed and reduce energy consumption in computing, starting with data centers.<\/p>\n

Backed by years of research at MIT and elsewhere, Ayar has developed chips that move data around with light but compute electronically. The unique design integrates speedy, efficient optical communications \u2014 with components that transmit data using light waves \u2014\u00a0into traditional computer chips, replacing less efficient copper wires.<\/p>\n

According to the startup, the chips can reduce energy usage by about 95 percent in chip-to-chip communications and increase bandwidth tenfold over their copper-based counterparts. In massive data centers \u2014 Ayar\u2019s first target application \u2014 run by tech giants such as Facebook and Amazon, the chips could cut total energy usage by 30 to 50 percent, says CEO Alex Wright-Gladstein MBA \u201915.<\/p>\n

\u201cRight now there\u2019s a bandwidth bottleneck in big data centers,\u201d says Wright-Gladstein, who co-founded Ayar with Chen Sun PhD \u201915 and Mark Wade, a University of Colorado graduate and former MIT researcher. \u201cThat\u2019s an exciting application and the first place that really needs this technology.\u201d<\/p>\n

In December, the startup penned a deal with GlobalFoundries, a top global silicon-chip manufacturer, to bring its first product,\u00a0an optical input-output system called Brilliant,\u00a0to market next year.<\/p>\n

The chips could also be used in supercomputers, Wright-Gladstein adds, which have similar efficiency issues and speed constraints as data centers do. Down the road, the technology could also improve optics in various fields, from autonomous vehicles and medical devices to augmented reality. \u201cWe\u2019re excited about not just what this can do for data centers, but what new things this will enable in the future,\u201d Wright-Gladstein says.<\/p>\n

Seeing the light<\/strong><\/p>\n

Ayar\u2019s core technology \u2014 now backed by more than 25 academic papers \u2014 is a decade in the making. The research collaboration began in the mid-2000s at MIT as part of the Defense Advanced Research Project Agency\u2019s Photonically Optimized Embedded Microprocessors (POEM) project, led by Vladimir Stojanovic, now an associate professor of electrical engineering and computer science at the University of California at Berkeley, in collaboration with Rajeev Ram, an MIT professor of electrical engineering and principal investigator for the Physical Optics and Electronics group, and\u00a0Milo\u0161 Popovi\u0107<\/span>, now an assistant professor of electrical and computer engineering at Boston University.<\/p>\n

The idea was to help data transmission keep up with Moore\u2019s Law. The number of transistors on a chip may double every two years, Wright-Gladstein says, \u201cbut the amount of data we push across those copper pins hasn\u2019t grown at the same rate.\u201d<\/p>\n

Computer chips send data between chips with different functions, such as logic chips and memory chips. With copper-based communications, however, the chips can\u2019t send and receive enough data to take advantage of their increasing processing power. That\u2019s caused a \u201cbottleneck,\u201d where chips must wait long durations to send and receive data. More than half the time in data centers, for instance, circuits are waiting for data to come and go, Wright-Gladstein says. \u201cIt\u2019s a huge waste,\u201d she says. \u201cThey\u2019re using almost as much power idling as when they are working.\u201d<\/p>\n

One solution is light. An optical wire can transmit multiple data signals on different wavelengths of light, while copper wires are limited to one signal per wire. Optical chips can, therefore, transmit more information using significantly less space. Moreover, photonics produce very little waste heat. Data passing through copper wires generates large amounts of waste heat, which hurts efficiency in individual chips. This is an issue in data centers, where copper wires run inside and between servers.<\/p>\n

At the time that the research groups of Ram, Stojanovic, and Popovic were working on the POEM project, large companies such as Intel and IBM were trying to design inexpensive, scalable optical chips. The collaboration \u2014 which then included Sun and Wade \u2014 took a different approach: They integrated optical components onto silicon chips, which are fabricated using the traditional CMOS semiconductor manufacturing process that churns out chips for pennies. \u201cThat was radical idea at time,\u201d Wright-Gladstein says. \u201cCMOS doesn\u2019t lend itself well to optics, so industry veterans assumed you\u2019d have to make major changes to get it to work.\u201d<\/p>\n

To avoid making changes to the CMOS process, the researchers focused on a new class of miniaturized optical components, including photodetectors, light modulators, waveguides, and optical filters that encode data on different wavelengths of light, and then transmit and decode it. They essentially \u201chacked\u201d the traditional method for silicon chip design, using layers intended for electronics to build optical devices, and enabling chip designs to include optics more tightly configured than ever inside a chip\u2019s structure.<\/p>\n

In 2015, the researchers, together with Krste Asanovic\u2019s team at UC Berkeley made the first processor to communicate using light and\u00a0<\/span>published<\/a>\u00a0<\/span>the results in\u00a0<\/span>Nature<\/em>. The chips, manufactured at a GlobalFoundries fabrication facility, contained 850 optical components and 70 million transistors, and performed as well as traditional chips manufactured at the same facility.<\/p>\n

Taking the plunge<\/strong><\/p>\n

\"The
The co-founders of Ayar Labs (from left) Chen Sun, Alex Wright-Gladstein, and Mark Wade.<\/figcaption><\/figure>\n

Behind the scenes, Wright-Gladstein was already thinking about commercialization. The year before the publication, she had enrolled in the MIT Sloan School of Management, specifically to meet researchers tackling clean energy. Taking 15.366 (Energy Ventures), which focuses on commercializing MIT clean technologies, she was chosen to select the technologies to bring into the classroom. \u201cThat was the perfect excuse to meet every researcher doing energy-related research,\u201d Wright-Gladstein says.<\/p>\n

From the vast pool of 300 labs, she came across Ram\u2019s optoelectronic chips \u2014 which \u201cblew me away,\u201d she says. The energy industry was focused on equipment innovations to save energy in data centers. \u201cBut there wasn\u2019t much focus on reducing energy through computing itself,\u201d Wright-Gladstein says. \u201cIt seemed like a great way to make an impact.\u201d<\/p>\n

Wright-Gladstein formed a team in class to create a business plan and pitch deck. She also collaborated frequently with Sun and Wade in speaking with potential industry clients. When the MIT Clean Energy Prize rolled around, the three students entered the technology under the name, OptiBit \u2014\u00a0and\u00a0<\/span>won both grand prizes<\/a>\u00a0<\/span>for $275,000, solidifying their decision to launch a startup.<\/p>\n

\u201cHaving funds early on to pay ourselves low salaries and having a little cushion before raising venture capital funds really persuaded us all to take the plunge,\u201d Wright-Gladstein says.<\/p>\n

Setting up shop in San Francisco, the startup continued research and development, increasing the communication data rates of the technology. Last year, GlobalFoundries took interest in these constant innovations and struck a partnership with the startup, which included some undisclosed funding. This year, Ayar\u2019s first prototypes should reach U.S. data centers, with a planned 2019 commercial release.<\/p>\n

Solving the chip input-output problem is just the start. Ayar is also excited about what its new technology means for the field of optics, Wright-Gladstein says. Optical sensors, for instance, are used in self-driving or semiautonomous vehicles and expensive medical equipment. Lowering manufacturing costs, while increasing the computation power, of optoelectronic chips could make those technologies much less expensive and more accessible.<\/p>\n

\u201cWe\u2019re starting out solving this bottleneck problem in traditional silicon chips, but ultimately we\u2019re excited about all the different places this technology will go,\u201d Wright-Gladstein says. \u201cThis is going to change the availability of optics, and how the world can use optics, in ways beyond what we can predict right now.\u201d<\/p>\n","protected":false},"excerpt":{"rendered":"

Startup\u2019s optoelectronic chips could reduce energy usage by up to 50 percent in data centers while increasing computing speeds. By Rob Matheson – MIT News Office, originally published on\u00a0MIT News. ECE Professor Milos Popovic moved to BU in mid-2016 from a previous position as Assistant Professor at University of Colorado Boulder. While at CU Boulder, […]<\/p>\n","protected":false},"author":14405,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[236,907,1119],"tags":[436,437,260,438,439,405,440,441,391,442,443,444,410,445,446,447,448],"_links":{"self":[{"href":"https:\/\/www.bu.edu\/eng\/wp-json\/wp\/v2\/posts\/69475"}],"collection":[{"href":"https:\/\/www.bu.edu\/eng\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.bu.edu\/eng\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.bu.edu\/eng\/wp-json\/wp\/v2\/users\/14405"}],"replies":[{"embeddable":true,"href":"https:\/\/www.bu.edu\/eng\/wp-json\/wp\/v2\/comments?post=69475"}],"version-history":[{"count":1,"href":"https:\/\/www.bu.edu\/eng\/wp-json\/wp\/v2\/posts\/69475\/revisions"}],"predecessor-version":[{"id":131229,"href":"https:\/\/www.bu.edu\/eng\/wp-json\/wp\/v2\/posts\/69475\/revisions\/131229"}],"wp:attachment":[{"href":"https:\/\/www.bu.edu\/eng\/wp-json\/wp\/v2\/media?parent=69475"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.bu.edu\/eng\/wp-json\/wp\/v2\/categories?post=69475"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.bu.edu\/eng\/wp-json\/wp\/v2\/tags?post=69475"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}