The huge increase in computing performance in recent decades has been achieved by squeezing ever more transistors into a tighter space on microchips leading to leakage effects resulting in delays known as the “interconnect bottleneck,” is becoming an increasing problem in high-speed computing systems.
While the computers of today use transistors and semiconductors to control electricity, the Optical Computers of the future may utilize crystals and metamaterials to control light particles called photons. Optical computers promise to be superfast since light travels at 186,000 miles per second. In a billionth of a second, one nanosecond, photons of light travel just a bit less than a foot, not considering resistance in air or of an optical fiber strand or thin film. Electronic computers are relatively slow, and the faster we make them the more power they consume,the future optical computers shall also be energy efficient.
Research teams are also making breakthroughs in the development of photonic computer chips that imitate the way the brain’s synapses operate. This would enable them to execute current machine learning and deep neural learning algorithms with high speed and energy efficient way.
The technology that is enabling optical processors and optical computers is silicon photonics. Silicon photonics uses photons to detect process and transmit information more efficiently than electrical signals, and yet have low manufacturing costs as a result of using conventional silicon-integrated-circuit processes. Silicon photonics refers to the application of photonic systems using silicon as an optical medium. The silicon material used in such photonic systems is designed with sub micrometer precision and is deployed into the microphotonic components. Silicon photonics combines technologies such as complementary metal oxide semiconductor (CMOS), micro-electro-mechanical systems (MEMS) and 3D Stacking.
Electronics-Photonics Hybrid computers
“Entirely optical computers are still some time in the future,” says Dr. Frazier, “but electro-optical hybrids have been possible since 1978, when it was learned that photons can respond to electrons through media such as lithium niobate. Newer advances have produced a variety of thin films and optical fibers that make optical interconnections and devices practical.
We are focusing on thin films made of organic molecules, which are more light sensitive than inorganics. Organics can perform functions such as switching, signal processing and frequency doubling using less power than inorganics. Inorganics such as silicon used with organic materials let us use both photons and electrons in current hybrid systems, which will eventually lead to all-optical computer systems.
“What we are accomplishing in the lab today will result in development of super-fast, super-miniaturized, super-lightweight and lower cost optical computing and optical communication devices and systems,” Frazier explained.
Single-chip microprocessor that communicates directly using light
Group of researchers at MIT, the University of California, Berkeley, and the University of Colorado, Boulder have developed an electronic-optical microprocessor by integrating over 70 million transistors and 850 optical components on a single chip working together to provide logic, memory, and interconnect functions, as described in the journal Nature.
The system uses on-chip photonic devices to directly communicate with other chips using light. “This is a milestone. It’s the first processor that can use light to communicate with the external world,” said Vladimir Stojanovic, a professor at the University of California, Berkeley, who led development of the chip. “No other processor has the photonic I/O in the chip.”
To integrate electronics and photonics at the scale of a microprocessor chip, we adopt a ‘zero-change’ approach to the integration of photonics. Instead of developing a custom process to enable the fabrication of photonics, which would complicate or eliminate the possibility of integration with state-of-the-art transistors at large scale and at high yield, we design optical devices using a standard microelectronics foundry process that is used for modern microprocessors.
The data transfers in the prototype occurred at a rate of 300 gigabits per second per square millimeter, which the researchers say is 10 to 50 times the rate for a comparable off-the-shelf electronic microprocessor. That boost in bandwidth could save a lot of energy in data centers, says Chen Sun, a researcher at the University of California, Berkeley.
He estimates that 20 to 30 percent of the energy used in data-center servers is spent transferring data between processor, memory, and networking cards. According to an analysis by the Natural Resources Defense Council, data centers in the United States will consume 140 billion kilowatt-hours of electricity a year by 2020, costing $13 billion and emitting 100 million metric tons of carbon.
This demonstration could represent the beginning of an era of chip-scale electronic–photonic systems with the potential to transform computing system architectures, enabling more powerful computers, from network infrastructure to data centres and supercomputers.
Silicon Photonic Neuromorphic chips and Neural Network
The research team has made the pioneering breakthrough of the development of photonic computer chips that imitate the way the brain’s synapses operate. The work, conducted by researchers from Oxford, Münster and Exeter universities, combined phase-change materials – commonly found in household items such re-writable optical discs – with specially designed integrated photonic circuits to deliver a biological-like synaptic response. Crucially, their photonic synapses can operate at speeds a thousand times faster than those of the human brain.
The PCM’s ability to absorb light changes when heated, which can be used to control the amount of light that passes through the waveguide. In previous research, the group had shown that optical pulses could be used to switch between various states of absorption to store information—effectively creating a photonic memory device.
The team believes that the research could pave the way for a new age of computing, where machines work and think in a similar way to the human brain, while at the same time exploiting the speed and power efficiency of photonic systems.
Professor C David Wright, co-author from the University of Exeter, said: ‘Electronic computers are relatively slow, and the faster we make them the more power they consume. Conventional computers are also pretty “dumb”, with none of the in-built learning and parallel processing capabilities of the human brain. We tackle both of these issues here – by developing not only new brain-like computer architectures, but also by working in the optical domain to leverage the huge speed and power advantages of the upcoming silicon photonics revolution.’
Professor Wolfram Pernice, a co-author of the paper from the University of Münster, added: ‘Since synapses outnumber neurons in the brain by around 10,000 to one, any brain-like computer needs to be able to replicate some form of synaptic mimic. That is what we have done here.’
Alexander Tait and pals at Princeton University in New Jersey have built an integrated silicon photonic neuromorphic chip and show that it computes at ultrafast speeds. “Photonic neural networks leveraging silicon photonic platforms could access new regimes of ultrafast information processing for radio, control, and scientific computing,” say Tait and co.
The authors have reported the first experimental demonstration of an integrated photonic neural network that also makes first use of electro-optic modulators as photonic neurons.The nodes take the form of tiny circular waveguides carved into a silicon substrate in which light can circulate. When released this light then modulates the output of a laser working at threshold, a regime in which small changes in the incoming light have a dramatic impact on the laser’s output.
A silicon-compatible photonic neural networking architecture called “broadcast-and-weight” has been proposed. In this architecture, each node’s output is assigned a unique wavelength carrier that is wavelength division multiplexed (WDM) and broadcast to other nodes. Incoming WDM signals are weighted by reconfigurable, continuous-valued filters called microring (MRR) weight banks and then summed by total power detection. This electrical weighted sum then modulates the corresponding WDM channel. A nonlinear electro-optic transfer function, such as a laser at threshold or, in this work, a saturated modulator, provides the nonlinearity required for neuron functionality.
They go on to demonstrate how this can be done using a network consisting of 49 photonic nodes. They use this photonic neural network to solve the mathematical problem of emulating a certain kind of differential equation and compare it to an ordinary central processing unit.
The results show just how fast photonic neural nets can be. “The effective hardware acceleration factor of the photonic neural network is estimated to be 1,960 × in this task,” say Tait and co. That’s a speed up of three orders of magnitude. “Silicon photonic neural networks could represent first forays into a broader class of silicon photonic systems for scalable information processing,” say Taif and co.
All Optical Processing
All-optical signal processing has been considered a solution to overcome the bandwidth and speed limitations imposed by conventional electronic-based systems. Over the last few years, an impressive range of all-optical signal processors have been proposed, but few of them come with reconfigurability, a feature highly needed for practical signal processing applications. Following the electronic component design strategies, many equivalent photonic signal processors, such as temporal differentiators, temporal integrators, Hilbert transformers and ordinary differential equation (ODE) solvers, have been proposed and demonstrated.
Two very relevant examples of these fundamental devices are temporal differentiators and ODE solvers. Temporal differentiators can be used to perform real-time differentiation of an optical signal in the optical domain and have been applied to ultrafast signal generation and pulse characterizations. ODE solvers play an irreplaceable role in virtually any field of science or engineering, such as automatic control and temperature diffusion processes, writes Ming Li from Institute of Semiconductors, Chinese Academy of Sciences, No. 35, Tsinghua East Road, Beijing, 100083, China, and others.
Andrea Blanco-Redondo and Dr Chad Husko from CUDOS (ARC Centre of Excellence for Ultrahigh bandwidth Devices for Optical Systems) at the University of Sydney’s School of Physics have observed an on-chip soliton compression in a silicon photonic crystal for the first time. Solitons are nonlinear waves that propagate through a medium undistorted and their discovery shall enable development of nonlinear devices in silicon and all-optical processing systems.
UK-based startup Optalysys promises optical processors for supercomputers
Optalysys has completed 320 gigaFLOP optical computer prototype, targets 9 petaFLOP product in 2017 and 17 exaFLOPS machine by 2020 .
“Optalysys’ technology applies the principles of diffractive and Fourier optics to calculate the same processor intensive mathematical functions used in CFD (Computational Fluid Dynamics) and pattern recognition,” Dr. Nick New, CEO and founder of the company explained. “Using low power lasers and high resolution liquid crystal micro-displays, calculations are performed in parallel at the speed of light,” New further described.
The Optalysys Optical Solver Supercomputer reduces energy footprint so it can be considered as eco-efficient as well. No need for special power as it only needs a standard mains power supply.
As for the running cost, the super processor only costs £2,100 every year. This is way cheaper than the fast supercomputer available in China, the Tianhe-2, developed by the National University of Defense Technology. The computer alone costs $320 million but with a $21m annual running cost.
Programmable optical chip
Photonic chip functionality is usually hard-wired by design, however reconfigurable optical elements would allow light to be routed flexibly, opening up new applications in programmable photonic circuits. The ability to control light in a silicon chip could enable novel applications in programmable photonic circuits, including applications for optical testing and data communication.
A field-programmable gate array (FPGA) is an integrated circuit designed to be programmed or configured by a customer or a designer after manufacturing – hence “field-programmable”— be it for speech recognition, computer vision, cryptography, or something else. In the heart of an FPGA is a large array of logic blocks that are wired up by reconfigurable interconnects, allowing the chip to be reconfigured or programmed via specialized software.
Improvements in both silicon photonics and III–V compound semiconductor technology, such as InP and GaAs are now enabling the development of optical equivalent of an FPGA. Researchers are starting to build designs of programmable optical signal processors on a chip by cascading arrays of coupled waveguide structures that feature phase shifters to control the flow of light through the array and thus support reconfigurability.
Reconfigurable Silicon Photonic Circuits Provide Control of Light Patterns
Traditional spatial light modulators are based on liquid crystals or micromirrors and provide many independently controllable pixels. This technology has revolutionised optics in recent years, with many applications in imaging and holography, adaptive optics and wavefront shaping of light through opaque media. Researchers at the University of Southampton and the Institut d’Optique in Bordeaux, France, multimode interference (MMI) devices, which form a versatile class of integrated optical elements routinely used for splitting and recombining different signals on a chip. The geometry of the MMI predefines its characteristics at the fabrication stage.
They demonstrated that light could be routed between the ports of a multimode interference (MMI) power splitter with more than 97 percent total efficiency and negligible losses. The intricate interplay between many modes traveling through the MMI was dynamically controlled. A pattern of local perturbations, induced by femtosecond laser, was used to shape the transmitted light, demonstrating that all-optical wavefront shaping in integrated silicon-on-insulator photonics devices is possible.
By employing UV pulsed laser excitation to modify the spatial refractive index profile, the research team was able to maintain control of the optical transfer of telecommunication-wavelength light traveling through the device, thus allowing the functionality of the light to be redefined.
Photonics chip functionality is typically hardwired. Reconfigurable optical elements that would provide the ability to freely route light in a static silicon element offer an important building block for field-programmable photonics, the researchers said. “We have demonstrated a very general approach to beam shaping on a chip that provides a wide range of useful functionalities to integrated circuits,” said research fellow Roman Bruck. “The integrated spatial light modulator turns conventional silicon photonics components into versatile reconfigurable elements”.