#OFC24: Moving Data at the Speed of AI

 Check out OFC Conference and Exposition 2024 videos here: https://ngi.fyi/ofc24yt

Ayar Labs is unveiling the second generation of SuperNova, which powers the company’s TeraPHY optical I/O chiplet and enables huge leaps in bi-directional bandwidth. Mark Wade, CEO and Co-founder from Ayar Labs, discusses:

- The unveiling of Ayar Labs' latest hardware, achieving full bandwidth between two chips of 4 terabits per second.

- The power of this technology to transmit massive amounts of bandwidth with low latency, high bandwidth densities, and excellent power efficiency, all while maintaining signal integrity.

- The ongoing ecosystem work, including optically connected FPGAs with Intel, high-density optical connector work with Teramount, and the management of optical connections with Corning around glass interposers and optical interposers with glass shuffles.

https://youtu.be/AEP2_pZQDrY

Want to be involved our video series? Contact info@nextgeninfra.io

Intel and Ayar Labs demo 4 Tbps optically-connected FPGA

Monday, August 28, 2023    

Intel and Ayar Labs are showcasing a 4 Tbps optical FPGA with Ayar Labs’ in-package optical I/O at this week’s HOT CHIPS 2023 event at Stanford Univertsity.The demo shows a 4 Tbps bidirectional communication between two optically enabled Intel FPGAs using Ayar Labs TeraPHY optical I/O chiplets. Data generated from each FPGA is optically transmitted from one board and received and verified on the other board.https://ayarlabs.com/hot-chi...

READ MORE

false

Read More

#OFC24: Big Power Savings with 800G Linear Receive Optics

Energy consumption is one of the biggest challenges with GenAI, so reducing the power consumed by optical modules, while increasing data rates, is a top priority. Bill Brennan, President and CEO from Credo, shares some insights:

- He discusses the importance of connectivity and the shift towards faster lane rates and denser connections in the face of AI advancements.

- Brennan introduces the innovative solution of Linear Receive Optics, which effectively halves the power of the DSP function in optical modules while maintaining industry standards.

- He highlights the significant energy savings achieved with this approach, allowing for more energy to be redirected towards compute, thereby enhancing the value of the cluster.

https://youtu.be/GhIq9Qv08bY

Want to be involved our video series? Contact info@nextgeninfra.io

false

Read More

SK hynix to build $3.8B advanced packaging fab in Indiana

SK hynix plans to invest an estimated $3.87 billion to build a state-of-the-art advanced packaging fabrication and R&D facility for AI products in West Lafayette, Indiana.

SK hynix is the world's leading producer of High-Bandwidth Memory (HBM) chips. The new facility will be home to an advanced semiconductor production line that will mass-produce next-generation HBM, the highest-performing Dynamic Random Access Memory (DRAM) chips, which are the critical components of graphic processing units that train AI systems such as ChatGPT. The company plans to begin mass production in the second half of 2028, while the new facility will also develop future generations of chips and house an advanced packaging R&D line.

At the event, officials from each party including Governor of Indiana Eric Holcomb, Senator Todd Young, Director of the White House Office of Science and Technology Policy Arati Prabhakar, Assistant Secretary of Commerce Arun Venkataraman, Secretary of Commerce State of Indiana David Rosenberg, Purdue University President Mung Chiang, Chairman of Purdue Research Foundation Mitch Daniels, Mayor of city of West Lafayette Erin Easter, Ambassador of the Republic of Korea to the United States Hyundong Cho, Consul General of the Republic of Korea in Chicago Junghan Kim, SK vice chairman Jeong Joon Yu, SK hynix CEO Kwak Noh-Jung and SK hynix Head of Package & Test Choi Woojin, participated.

"SK hynix is the global pioneer and dominant market leader in memory chips for AI," Purdue University President Mung Chiang said. "This transformational investment reflects our state and university's tremendous strength in semiconductors, hardware AI, and hard tech corridor. It is also a monumental moment for completing the supply chain of digital economy in our country through chips advanced packaging. Located at Purdue Research Park, the largest facility of its kind at a U.S. university will grow and succeed through innovation."

"We are excited to become the first in the industry to build a state-of-the-art advanced packaging facility for AI products in the United States that will help strengthen supply-chain resilience and develop a local semiconductor ecosystem," said SK hynix CEO Kwak Noh-Jung. "With this new facility, we hope to advance our goal of providing AI memory chips with unmatched capabilities, serving the needs of our customers."

SK hynix is collaborating with Purdue University, one of the leading research institutions in the U.S., on plans for future R&D projects. These projects include work on advanced packaging and heterogeneous integration with Purdue's Birck Nanotechnology Center and other research institutes and industry partners. They also hope to collaborate on a project related to memory-centric solutions and architecture for the generative AI era, specifically system-level memory design and in/near-memory computing.

The company also plans to collaborate with Purdue University and Ivy Tech Community College to develop training programs and interdisciplinary degree curricula that will cultivate a high-tech workforce and build a reliable pipeline of new talent.

Separately, SK hynix will also proceed with Korean domestic investments as planned. The company has been working to prepare the site for the Yongin Semiconductor Cluster where it will invest 120 trillion won to build production facilities. The company plans to break ground on the first fab in March of 2025, with plans for completion in early 2027. It will also build a mini fab, a facility with equipment for 300mm wafer processing, to test semiconductor materials, components and equipment.

https://news.skhynix.com/sk-hynix-signs-investment-agreement-of-advanced-chip-packaging-with-indiana/

false

Read More

Zayo expands its nationwide 400G backbone

Zayo announced its latest round of critical infrastructure upgrades, including its newest long-haul dark fiber route connecting Columbus to Pittsburgh and a highly sought-after 400G-enabled wavelength route connecting Washington, D.C. and Atlanta. The new long-haul dark fiber route connecting Columbus to Pittsburgh is carrier-diverse, 100% underground, and built with a high fiber count for maximum scalability and uptime availability.

Zayo has also added four additional routes to its 400G network—currently the largest 400G network in North America—in key markets including:

• Indianapolis to Columbus
• Cut Bank, Mont. to Calgary
• New Orleans to Ponchatoula, La.
• Grand Rapids, Mich. to Lansing, Mich.

Zayo is also expanding its leading IP network with the addition of 27 critical IP points of presence (PoPs) and 400G peering locations across North America. The addition of new IP PoPs and peering locations to Zayo’s tier-1 IP network guarantees scalable and dependable capacity availability. Built on Zayo’s wholly-owned fiber backbone, Zayo’s IP network also ensures enhanced reliability and performance with reduced latency to support the high-speed connectivity these applications require.

“While technology innovations like AI and next-gen cloud surge ahead, continued investment in network infrastructure to support this demand is critical. As the largest independent network, Zayo is in a prime position to bring network infrastructure up to par with today’s innovation, and ensure we and our customers are ready for what’s next,” said Bill Long, Chief Product & Strategy Officer at Zayo. “Our latest infrastructure expansions demonstrate our continuous efforts to provide our customers with unique, diverse, low-latency routes — where they need it today and where we know they'll need it tomorrow.”

https://www.zayo.com

false

Read More

Casa Systems disbands under Chapter 11

Citing "a significant decline in revenue and profits due in large part to industry-wide downward capital investment and procurement trends in the cable and telco markets," Casa Systems filed  voluntary petitions for debt relief under Chapter 11 of the Bankruptcy Code in the U.S. Bankruptcy Court for the District of Delaware. 

As part of this process, the company agreed to sell its 5G Mobile Core and RAN businesses to Lumine Group and have asked the court for approval to complete the transaction by the end of April.

Additionally, Casa Systems secured a stalking horse bid from an affiliate of Vecima for our Cable and Virtual Broadband Network Gateway (vBNG) business, and anticipate that an auction process will follow, in which other parties will have the opportunity to submit bids for the business.

Casa Systems belives this sale process is the best and most viable path forward to maximize value, preserve jobs and minimize disruption for its customers.

https://www.casasaleprocess.com

false

Read More

AT&T commits $3 billion to closing Digital Divide

AT&T is committing an additional $3 billion by 2030 to help close the digital divide, bringing the company's total commitment to $5 billion since 2021. 

This cumulative commitment includes an ambitious goal to help 25 million people get and stay connected to affordable, high-speed internet access throughout this decade. Working to close the digital divide is the linchpin of AT&T's mission to leave no one behind when it comes to online opportunity. This commitment is a pillar of the work the company does to help build this country's digital infrastructure.

"Connectivity is critical for success and prosperity in America today. Whether it's a first-generation high school graduate applying for college, a veteran accessing telehealth services, or a grandparent connecting with their family thousands of miles away — we want to help unlock greater possibility for the millions of people who remain unconnected," said AT&T CEO John Stankey. "The pace of innovation and the global economy aren't waiting on anyone, and we're committed to bringing connectivity and digital literacy in reach for all Americans." 

Since 2021, nearly 5 million Americans have benefited from our discounted internet service and digital literacy programs.

AT&T continues to expand its network of Connected Learning Centers (CLC) in communities across the country that are equipped with high-speed internet connections and computers supplied by Dell.

Read More

#OFC24: 800G Linear Receive Optics Connected to 51.2T Switch

At #OFC24, Credo and Wistron demonstrated Linear Receive Optics (LRO), using the Credo Dove 850 DSP, with the 51.2Tb/s Wistron DS-6183-64O 64-port 800GbE switch.  Easwar Sankar, Optical FAE Director from Credo Semi, explains:

- The introduction of Credo's Linear Receive Optics (LRO) product, a response to industry demand for more efficient DSP solutions. This innovative product addresses the shortcomings recognized in previous LRO attempts by other vendors.

- The unique design of the LRO, which includes a retimer on the transmit sided and a linear receiver. This allows for improved performance of the switch 30s while addressing all the impairments on the transmitter. This design enables customers to tune the transmitter for different kinds of optics.

- The benefits of the LRO, including a significant reduction in power and cost. The LRO is practical, deployable in mass volume, and can be used in a significant percentage of the current DSP-based transceiver market. 

https://youtu.be/5FrEo76n9Co

In an LRO transceiver, or Active Optical Cable (AOC), only the transmit path from the electrical input to the optical line side output includes a DSP for signal retiming and equalization. This unlocks dramatic power savings without sacrificing interoperability or standards compliance. All three 800G OSFP LRO modules provided by Credo partners for this demonstration seamlessly “plug and play” with the 51.2T Wistron switch.

Want to be involved our video series? Contact info@nextgeninfra.io

false

Read More

#OFC24: Intel's Optical I/O Chiplet Co-Packaged with Server CPU

At #OFC24, Intel demonstrated an advanced Optical Compute Interconnect (OCI) chiplet co-packaged with a prototype of a next-generation Intel CPU.

The OCI chiplet contains a single Silicon Photonics Integrated Circuit (PIC) with integrated lasers, an electrical IC with RF Through-Silicon-Vias (TSV), and a path to incorporate a detachable/re-usable optical connector.  The 4 Tbps bidirectional chiplet is compatible with PCIe Gen5, supporting 64 lanes of 32 Gbps data in each direction over 10’s of meters, realized as eight fiber pairs each carrying eight DWDM wavelengths.

Christian Urricariet, Senior Director of Product Marketing, from Intel's Silicon Photonics Products Division explains:

- Intel demoed its first Optical Compute Interconnect(OCI) chiplet, a high-density optical interconnect solution for compute elements like CPU or GPU clusters.

- The capabilities of the OCI chiplet, which can support up to 4 terabits per second bi-directionally, and its integration with next-generation Intel CPUs.

- The performance of the OCI technology, showcasing its ability to run over standard, widely deployed single-mode fiber (SMF-28), without requiring Polarization Maintaining Fiber (PMF) like other technical approaches in the market.

https://youtu.be/xYmiAS5zoFE

See Intel's blog on this topic:

https://community.intel.com/t5/Blogs/Tech-Innovation/Artificial-Intelligence-AI/Intel-Shows-OCI-Optical-I-O-Chiplet-Co-packaged-with-CPU-at/post/1582541

Want to be involved our video series? Contact info@nextgeninfra.io

false

Read More

#OFC24: Coherent Pluggables for 25GE to 400GE Metro Aggregation

 Check out OFC Conference and Exposition 2024 videos here: https://ngi.fyi/ofc24yt

DZS is working with Infinera to bring game-changing new integrated ICE-X Coherent Pluggable solutions to market for mobile xHaul and ultra high-speed symmetrical business services from 25GE to 400GE. The solution combines the recently introduced DZS Saber 2200 Coherent Ethernet Switch and Infinera’s intelligent coherent pluggables. Jason Lauzon, Strategic Sales Director at DZS, explains:  

- The introduction of the DZS Saber 2200, a high-capacity layer 2 Ethernet aggregation switch that can act as an XR Leaf node

- The unique features of Saber 2200, including 24 ports of 1, 2, 10, and 25 gig with four coherent line ports that can be XR Optics acting as a leaf note or 100 Gig coherent using QSFP DD technology.

- The DZS Saber 2200 will operate as a leaf node leveraging Infinera’s ICE-X’s ability to support point-to-point or point-to-multi-point applications. The integrated solution enables service providers to cost-efficiently deliver 25GE, 50GE, 75GE, 100GE and 200GE enterprise-class services over current Passive Optical Networking (PON) infrastructure. 

https://youtu.be/RXyKg_C7jcs

Solution highlights:

• High capacity, low latency comprehensive Layer 2 networking support enabling mobile front/mid/backhaul capabilities
• Key Time Sensitive Networking features including Integrated Grand Master Timing
• Supports enhanced common public radio interface eCPRI/CPRI transport capabilities via Radio Over Ethernet (ROE)
• Advanced Flexible Ethernet (FlexE) support enables service providers to optimize bandwidth and improve network efficiencies
• Supports point-to-multipoint applications over 100GE interfaces
• Enables the ability to dramatically reduce the number of optical transceivers and aggregation sites, offering significant CapEx reductions
• Enables service providers to maximize router efficiency, density, and simplicity, while aligning CapEx and actual bandwidth requirements
• Seamlessly adapts to changing bandwidth demands and traffic patterns
• NETCONF/YANG managed
• Supports redundant, field-replaceable power units

Stay tuned for more updates and technology use cases from DZS and their partnership with Infinera.

Want to be involved our video series? Contact info@nextgeninfra.io

false

Read More

#OFC24: Validating 800G links and AI/ML Workloads

 There is an acute need to emulate AI/ML workloads in order to validate the performance of network infrastructure prior to a major AI infrastructure deployment. Asim Rasheed, Manager from Spirent explains:

- Showcasing Spirent's range of products, from 100G to 800G, including the M1 Compact Appliance with 16 ports of 400G.

- The Spirent A2 400G Appliance is a flexible platform with high-density 400G ports for mission-critical scale emulation and select functional testing, such as throughput testing with traffic packet blasting, and proof of concept (PoC) labs to emulate a real-world service provider customer network. 

- The next-generation 2U platform is available in 8-port and 16-port variants for high-density 400G testing and protocol scale emulation, such as high-density hardware interoperability testing and ASICs testing for chipset vendors.

- The demonstration of the Juniper switch, handling 1.6 Tbps of traffic via 2 800G ports of live traffic, running at line rate.

https://youtu.be/CDkGCF2oK6M

Want to be involved our video series? Contact info@nextgeninfra.io

false

Read More

#OFC24: Spirent's 800G 802.3df Appliance & LPO optics support

Check out OFC Conference and Exposition 2024 videos here: https://ngi.fyi/ofc24yt

Test and measurement solutions by necessity must be ahead of vendor solutions. So what's next in switching? Steve Rumsby, Senior Director of Architecture & Platform at Spirent, provides highlights of Spirent's OFC24 demonstrations:

- The introduction of Spirent's B3 800G 802.3df Appliance offering 800G Ethernet in the latest IEEE 802.3df configuration, which includes Auto negotiation Link Training and a wide range of Ethernet topologies.

- The showcase of various interconnect solutions at the Spirent booth, including active electrical cables, active optical cables, LPO technology, and traditional transceivers, all operating at 800G

- The benefits and functionality of LPO (Linear Pluggable Optics) technology, an interconnect that converts electrical signals directly to optical without a DSP, resulting in lower power and lower latency. 

https://youtu.be/1a2XqewNmHo

Want to be involved our video series? Contact info@nextgeninfra.io

false

Read More

#OFC24: AMD's 800G Progress & Multi-Layer Security

Check out OFC Conference and Exposition 2024 videos here: https://ngi.fyi/ofc24yt

What were the big trends at this year's #OFC24? Gilles Garcia, Senior Director, Business Lead from AMD, provides an insightful overview:

- The emergence of 800 gig technology: Garcia discusses the importance of 800 gig, its role in infrastructure, and its future prospects. He also highlights AMD's 800G FPGA-based solution, demonstrating its interoperability and maturity in the current ecosystem.

- The increasing importance of security in infrastructure: Garcia emphasizes the need for security at every layer within the transport networks. He explains how AMD is addressing this need with three types of security solutions.

- The future of security with post-quantum cryptography: Garcia delves into the future of security, discussing the potential impact of quantum computing on current cryptography algorithms. He presents AMD's post-quantum crypto capable systems as a solution for safe key exchange in the future.

https://youtu.be/2lnP-32sRA4

Join us as we explore these exciting developments in wireless technology and security.

Want to be involved our video series? Contact info@nextgeninfra.io

false

Read More

Digital Realty to Host NVIDIA AI Supercomputer in Denmark

Digital Realty has been selected to host one of the world's most powerful AI supercomputers at an AI-ready data center in Copenhagen, Denmark. 

The Novo Nordisk Foundation and the Export and Investment Fund of Denmark (EIFO) are joining forces to provide Danish researchers from both the public and private sectors with access to an NVIDIA-powered AI supercomputer. This initiative will not only offer them the supercomputer's capabilities but also access to NVIDIA's software platforms, training, and expertise.

This advanced NVIDIA DGX SuperPOD supercomputer boasts 191 NVIDIA DGX H100 systems, which incorporate more than 1,500 NVIDIA H100 Tensor Core GPUs. These are linked together using NVIDIA's Quantum-2 InfiniBand networking technology. Additionally, the supercomputer comes with NVIDIA AI Enterprise software, which includes pretrained models, optimized computing frameworks, and libraries for faster data science. Housed within Digital Realty's state-of-the-art, AI-optimized infrastructure, this powerful setup empowers the newly established National Center for AI Innovation in Denmark. This center aims to push forward research and innovation across various sectors such as healthcare, life sciences, and climate studies.

The supercomputer will be housed in one of Digital Realty's state-of-the-art data centers in Copenhagen, which have been powered by 100% renewable energy since 2020. Digital Realty has more than one gigawatt of renewable energy under contract globally and is a signatory to the EU Climate Neutral Data Centre Pact, committing to carbon neutrality for its European data centers by 2030.

Chris Sharp, Chief Technology Officer at Digital Realty, commented, "Digital Realty is at the forefront of crafting the digital future, transcending mere infrastructure development to become architects of tomorrow. Our commitment to delivering state-of-the-art, AI-ready infrastructure is not just about accelerating AI adoption – it's about pioneering the next wave of innovation. We envision Digital Realty as more than a meeting place for AI; we aim to establish it as the premier nexus for private AI, where enterprises and data converge. Here, scientific breakthroughs are realized, industries are transformed, and the future is molded by the synergy of technology and imagination."

"We are delighted to be working with Digital Realty," said Mads Krogsgaard Thomsen, Chief Executive Officer of the Novo Nordisk Foundation. "With their superb track record of operational excellence and tangible commitment to sustainability, they are the ideal partner to host the supercomputer and help take this exciting initiative forward."

Read More

Dell'Oro: Campus NaaS market to double in 2024

Dell'Oro Group projects that Campus Network as a Service (CNaaS) revenues will double in 2024 in the midst of a LAN equipment market contraction.

"Campus NaaS uses cloud-managed equipment to simplify LAN connectivity for enterprises," said Siân Morgan, Wireless LAN Research Director at Dell'Oro Group. "CNaaS vendors are changing the way companies consume Wi-Fi in their network, and we expect this model of service to grow faster than the overall LAN market for the next five years. CNaaS startups such as Nile, Join Digital, and Shasta Cloud have begun partnering with service providers to extend the reach of their innovative offers.

"Public Cloud-Managed LAN equipment has always grown faster than the overall LAN equipment market, but that growth differential is narrowing. The next wave of high growth in the LAN equipment market will come from CNaaS. We've identified 13 vendors with CNaaS, and with the new CNaaS Service Packs that HPE has released, the company has the largest breadth of available offers," add Morgan.

Additional highlights from the Campus NaaS and Public Cloud-Managed LAN Advanced Research Report:

• On average, unit pricing for LAN equipment is forecasted to decline over the next five years now that lead times have returned to normal.
• Recurring license revenues outgrew expectations in 2023, promising to contribute considerable revenues to both Public Cloud-Managed LAN and to CNaaS.
• The market has converged on three types of Campus NaaS offers: Enabler, Turnkey, and LAN-as-a-Utility. Each type of Campus NaaS has different characteristics and growth drivers.
• The CNaaS model is expected to flourish in two particular use cases which benefit from a recurring cost structure and a fully-managed network.

https://www.delloro.com/news/campus-naas-and-public-cloud-managed-lan-revenues-to-exceed-12-b-in-2028/

Read More

#OFC24: A New Class of Silicon Photonics for AI Data Centers

Silicon Photonics is poised to play a much greater role in data centers going forward. 

Loi Nguyen, EVP and GM, Optical from Marvell explains how:

- The introduction of Marvell's  Marvell 3D SiPho Engine, a highly integrated silicon photonics engine that can deliver 6.4 Tbps, enabling faster speeds for future AI technologies.

- The successful use of silicon photonics technology in the metro market to connect data centers, and its potential for wider use within data centers.

- The increasing demand for bandwidth due to AI and machine learning, and how silicon photonics can meet this demand with highly integrated optics.

Featuring 200G I/O, 3D SiPho Engine delivers 2x the bandwidth, 2x the input/output (I/O) bandwidth density, and 30% lower power per bit versus comparable devices with 100 Gbps electrical and optical interfaces.

https://youtu.be/TC2-S66FrOM

Want to be involved our video series? Contact info@nextgeninfra.io

false

Read More

#OFC24: Indium Phosphide (InP) Interconnect for AI Infrastructure

 Check out OFC Conference and Exposition 2024 videos here: https://ngi.fyi/ofc24yt

Indium Phosphide (InP) components will play a critical role connecting AI clusters across and between data centers, yet only a few vendors have deep expertise in Indium Phosphide. Robert Shore, SVP of Marketing from Infinera, provides a quick primer:

- The challenges faced by network operators due to the rapid growth of artificial intelligence, particularly in building and interconnecting AI clusters in data centers.

- The limitations of existing technologies like VCSELs or silicon-based technologies in transmitting information over longer distances within data centers, and how Infinera's Indium Phosphide technology excels in this area.

- Infinera's integrated InP photonic integrated circuit not only lowers costs and increases manufacturability, but also reduces power per bit, a crucial factor for data center operators.

- Infinera can help network operators to establish a secure supply chain in the United States of highly-integrated, low-power, and high-performance optics for AI clusters.

https://youtu.be/koBOzg4SmjU

Want to be involved our video series? Contact info@nextgeninfra.io

false

Read More

#OFC24: Co-Package Optics (CPO) Takes a Huge Step Forward

 The big players are now making significant bets on co-packaged optics (CPO) as a driver a for the highest performance interconnects in AI clusters. Hamid Arabzadeh, CEO from Ranovus, gives us the update from #OFC24:

- The significant investment in co-packaged optics by major industry players like TSMC, Nvidia, Broadcom, AMD/Xilinx, SilTerra, and Mediatek, and how this technology is being utilized in AI and Edge AI applications where traditional modules don't fit.

- The development of a new generation of co-packaged optics that includes the laser inside the module, allowing for breakthroughs in power consumption and beachfront density where copper could not provide.

- The design and manufacturing process of these optics, integrating multiple components inside a single die, using the silicon photonics substrate as a packaging platform, and achieving substantially lower power consumption than other technologies.

https://youtu.be/cdcAgMlsSFY

Keep up to speed with CPO, as it is poised to have a major impact of AI system design.

Want to be involved our video series? Contact info@nextgeninfra.io

false

Read More

#OFC24: Demonstrating Marvell's Latest Portfolio Solutions for AI

Marvell is building a full-range of solutions for AI infrastructure. At #OFC24. George Hervey, Principal Architect from Marvell, shows a full-rack with the latest technologies:

- A demonstration of Marvell's racks of solutions for AI infrastructure, including their latest product portfolio and coherent optics driven by their integrated MacSec retimer solution.

- An exploration of their 51 terabit switch solution, showcasing interaction with their AEC cable solution at 800 gig, extending the reach up to 5m and 

- A detailed look at their 51T platform demonstrating a liquid cooling option developed by their partners in Taiwan, and an additional 51T platform using 64 800F OSFP modules, showcasing interoperability across the portfolio.

https://youtu.be/JX9B4TO6Vmg

Want to be involved our video series? Contact info@nextgeninfra.io

false

Read More

#OFC24: Transmit-Retimed Optical DSPs Yield Huge Power Savings

 A new category of optical DSPs, Marvell's new Spica Gen2-T is designed for transmit retimed optical (TRO) modules, an emerging class of modules for AI and cloud use cases. Transmit-only DSPs enable TROs to reduce power consumption by more than 40% compared to conventional transceivers. Xi Wang, Vice President from Marvell, provides the overview:

- Introduction of Marvell's new product, the Spica Gen2-T, a transmit-only DSP that enables a new class of optical module, the Transmit Real Time Optical module.

- The Transmit Real Time Optical module allows operators to save 40% of power consumption while maintaining flexibility and scalability with a DSP based module.

- With this new class of optical module, not only is performance guaranteed and power reduced, but operators are satisfied and we contribute to saving the world. 

https://youtu.be/SJBKArfJYSY

Want to be involved our video series? Contact info@nextgeninfra.io

false

Read More

#OFC24: New 800ZR and 800ZR+ Pluggables Extend Reach

 The coming wave of 800ZR and 800ZR+ pluggable optics deliver outstanding performance over distance.  Stephen Adolph, AVP, CDSP Product Line Management from Marvell explains:

- At #OFC24 Marvell demonstrated its Orion DSP housed inside an OSFP module, perfect for for hyperscale network operators who need more distance from their network.

- The introduction of the Probabilistic Constellation Shaping (PCS) technology. This technology is important as it allows customers to extend their 800G reach to over 1,000 kilometers.

- The progression of technology from 800ZR, which supports up to 100 kilometers, to 800ZR+, which supports up to 500 kilometers, and finally to PCS, which significantly extends the reach. This progression is key to keeping the cost of scaling networks down and extending their capabilities to higher reach and capacities.

https://youtu.be/kn7B_zkeEQI

Want to be involved our video series? Contact info@nextgeninfra.io

false

Read More