Wednesday, April 17, 2024

Video: Intel's Silicon Photonics Roadmap

Amit Nagra, VP and GM, Silicon Photonics Product Division from Intel, shares an update on progress in silicon photonics, wafer manufacturing, and 1.6T components following this year's OFC24:

- Intel's strategic shift in the pluggable market: Instead of providing the full transceiver module, Intel is now focusing on supplying the key chipsets that go into these modules. Interestingly, five out of the six key pieces of silicon that go into a module can be sourced internally from Intel.

- Investment in photonics wafer manufacturing technology: Intel continues to invest in this technology in their New Mexico Fab, developing a next-generation process that improves the size, capability, speeds, and features of the process. This investment also allows Intel to transform the economics to create very compelling and cost-competitive silicon photonic solutions.

- Announcement of 1.6 Tbps components: At OFC 2024, Intel announced 1.6 Tbps components, both on the transmit side and the receive side, supporting 200G/lane for both the DR and FR formats. Intel aims to have the same coverage of the five components, including the transmitter PIC, the receiver PIC, and the associated electronics IC (EIC) and controller IC.

Learn more at 

Want to be involved our video series? Contact

Check out OFC Conference and Exposition 2024 videos here:


Intel scales its neuromorphic system

Intel provided an update on its large-scale neuromorphic systems based on its human brain-inspired Loihi 2 processors.

A new large-scale system (code named Hala Point ), which is deployed at Sandia National Laboratories, achieves over 10 times more neuron capacity and up to 12 times higher performance compared with its first-generation large-scale research system (code named Pohoiki Springs).

Intel said its Hala Point represents a significant advancement in neuromorphic computing, a brain-inspired approach designed to enhance the performance and efficiency of artificial intelligence (AI) systems. As the first large-scale neuromorphic system, Hala Point has demonstrated state-of-the-art computational efficiencies in mainstream AI workloads, capable of performing up to 20 quadrillion operations per second (20 petaops). Its efficiency is particularly notable, achieving more than 15 trillion 8-bit operations per second per watt (TOPS/W) when running conventional deep neural networks. 

Sandia National Laboratories is set to utilize Hala Point forbrain-scale computing research. The focus will be on tackling complex scientific computing problems across various domains, including device physics, computer architecture, and informatics. This usage underscores the system's potential to address high-level scientific and engineering challenges, leveraging its unique capabilities to advance knowledge in these critical areas. 

Hala Point is integral to addressing sustainability challenges associated with the rapid scaling of deep learning models, which have grown to trillions of parameters. The recent developments in neuromorphic computing, including those showcased at the International Conference on Acoustics, Speech, and Signal Processing (ICASSP), emphasize the potential of this technology to revolutionize AI hardware.

Key Points:

  • High Performance: Hala Point achieves up to 20 petaops with an efficiency exceeding 15 TOPS/W, surpassing GPU and CPU architectures.
  • Research Applications: Sandia National Labs will utilize Hala Point for advanced scientific and computing research.
  • Sustainability Focus: Neuromorphic computing addresses AI's sustainability challenges by integrating memory and computing for energy efficiency.
  • Neuromorphic Advancements: Loihi 2 processors, the basis for Hala Point, utilize brain-inspired computing principles for significant energy and performance gains.
  • Industry Collaboration: Intel collaborates with over 200 entities in the INRC to further develop and commercialize neuromorphic computing technologies.

“The computing cost of today’s AI models is rising at unsustainable rates. The industry needs fundamentally new approaches capable of scaling. For that reason, we developed Hala Point, which combines deep learning efficiency with novel brain-inspired learning and optimization capabilities. We hope that research with Hala Point will advance the efficiency and adaptability of large-scale AI technology,” states Mike Davies, director of the Neuromorphic Computing Lab at Intel Labs.

“Working with Hala Point improves our Sandia team’s capability to solve computational and scientific modeling problems. Conducting research with a system of this size will allow us to keep pace with AI’s evolution in fields ranging from commercial to defense to basic science,” said Craig Vineyard, Hala Point team lead at Sandia National Laboratories.

Australia's nbn tests Nokia's next-gen PON

Australia's NBN Co. completed the world’s first live network demonstration of multiple next-generation PON technologies using Nokia's fiber access platform.

The tests demonstrated 10G, 25G, 50G and 100G broadband speeds over nbn's existing fiber network.

Nokia’s Lightspan platform is designed to support a full range of PON technologies, from GPON to 100G PON.

The company says the tests show how operators can easily enhance 10G PON to symmetrical 25G PON and eventually evolve to 50G PON or 100G using the same passive and active fiber components.

Dion Ljubanovic, nbn Chief Network Officer, said: “This is an incredible result and demonstrates the fiber we deploy in our network is only limited by the capabilities of the equipment that will connect to it into the future. We are always looking for ways to push the capabilities of the nbn network and with the average Australian home forecast to download around 1 terabyte per month within the next decade, this successful demonstration showcases the world-class capabilities of nbn fiber to support the next generation of broadband services in the long term.”

Geert Heyninck, Vice President of Broadband Networks at Nokia, said: “There is a huge opportunity for operators to leverage their existing fiber broadband networks to efficiently add advanced services which goes way beyond consumer services. Think enterprise, mobile backhaul, Smart City, Industry 4.0. It’s important for service providers to have choices to match the right speed and cost points to meet the different use cases. As the industry’s first and only solution capable of supporting the full range of PON technologies from 10G to 25G, 50G, and even 100G, we can give operators the freedom and flexibility to meet their business needs while also helping to optimize network performance and reduce costs.”

The PON Wars: Debunking the 25G vs. 50G Debate

by Maurizio Severi, Head of Fiber Business Line at Nokia Fixed NetworksWhen listening to the arguments in the telecom industry about 25G and 50G PON, you could get the impression that there’s some kind of PON war going on. Even though 25G PON is being deployed, fierce debate still surrounds it. However, the debate concerning 25G versus 50G PON is fruitless, and it detracts from what is important. It’s not about the technology; it is never about...


SiTime debuts MEMS Integrated Clock Chip for Data Centers

SiTime unveiled its Chorus family of clock generators for AI datacenter applications. 

The company says its Chorus  family of clock-system-on-a-chip (ClkSoC) devices achieves a tenfold increase in performance while occupying only half the size of traditional standalone oscillators and clocks. This innovative design integrates clock, oscillator, and resonator technologies into a single chip, streamlining system clock architecture and reducing design time by up to six weeks. By incorporating timing products recently acquired from Aura Semiconductor, Chorus enhances its strategy to provide a comprehensive suite of uniquely differentiated solutions.

Key Features of SiTime’s Chorus SiT91211 and SiT91213 Clock Generator

  • Higher Performance: Delivers up to 10X better resilience with integrated MEMS resonator.
  • Design Simplicity: Integrated MEMS technology speeds the design process and eliminates common issues such as noise and impedance matching with integrated MEMS resonator.
  • Smaller Footprint: up to half the size in a 4 mm x 4 mm QFN.
  • Low RMS Phase Jitter: 70 femtoseconds typical (12 kHz to 20 MHz).
  • Flexible Frequency: Programmable frequency from 1 MHz to 700 MHz.
  • Flexible Output Types: Up to four differential (LVPECL, LVDS, LPHCSL) or eight LVCMOS outputs.
  • Flexible Supply Voltage: Programmable, 1.8V, 2.5V, or 3.3V.
  • Reduced Power and Simplified Circuitry: FlexSwing™ output reduces power consumption and eliminates termination resistors.
  • Excellent Frequency Stability: ±20 ppm and ±50 ppm from -40°C to 105°C.
  • EMI Reduction: Configurable spread-spectrum clock generation.
  • Compliant with the Latest PCIe Standard: Generation 1 to 6.
  • Enhanced System Robustness: Clock fault monitors (Lock Loss).

“AI is driving tremendous needs for higher data throughput in datacenters and lower power consumption, and SiTime is uniquely positioned to help address these issues,” said Piyush Sevalia, executive vice president of marketing at SiTime. “Before Chorus, hardware designers had to use discrete product types, such as clocks, oscillators and resonators, which resulted in performance compromises. Chorus delivers integrated clock generators to solve these problems and is yet another example of how we are transforming the timing market with our unique approach.”


Colt Data Centre Services expands in West London

Colt Data Centre Services (Colt DCS) has acquired land next to its existing site at Hayes, West London, with the goal to make the campus one of the largest in the UK.

West London is a key development location for data centres spurred on by the wider adoption of the public cloud by UK businesses and is becoming one of the most prominent submarkets in Europe. London's standing as one of the world's leading financial, media and business hubs remains a draw for companies looking to grow in the UK and Western Europe. 

Colt DCS is rapidly developing its initial site in Hayes, which will feature two buildings and provide 60MW of IT power. The first phase of this site is set to be completed by the third quarter of 2025. DCS said it has already seen significant interest from existing customers wanting space in this location. CThe company plans to expand with a second site, more than doubling the capacity of the London Hayes Data Centre campus. This expansion will make it one of the UK's largest data center sites and the biggest multi-building campus in Colt DCS's global network. The company has secured land and renewable energy sources necessary to support this growth, ensuring they can meet the rising demand for high-density AI workloads.

Additionally, the new site will contribute to Colt DCS’s goal of promoting a sustainable hyperscale data center future and assist in scaling customer operations. It will also benefit the local community socially and economically through involvement in educational programs, job creation, and business development initiatives.

DCS currently operates 18 state-of-the-art-carrier neutral data centers spanning 8 cities.

Richard Wellbrock, Chief Commercial Officer, at Colt DCS, said: "There is continued strong demand from customers for UK-based hyperscale data centres, driven by continued growth in demand for digital services. To cater for the fast-paced world of digital services, customers require scale in their data centre solutions. We are also excited to learn about the UK Government's plans to take the digital economy to the next level in the global marketplace, and our expansion in Hayes will increase the UK's capacity and capability in creating a vibrant environment for growth. We are committed to create a lasting sustainable impact on the local communities near our facilities and remain a trusted partner for our customers worldwide."