SANTA CLARA, Calif. – May 30, 2024 – Marvell Technology, Inc. (NASDAQ: MRVL), a leader in data infrastructure semiconductor solutions, today expanded its connectivity portfolio with the launch of the new Alaska® P PCIe retimer product line built to scale data center compute fabrics inside accelerated servers, general-purpose servers, CXL systems and disaggregated infrastructure. The first two products, 8- and 16-lane PCIe Gen 6 retimers, connect AI accelerators, GPUs, CPUs and other components inside server systems.
Artificial intelligence (AI) and machine learning (ML) applications are driving data flows and connections inside server systems at significantly higher bandwidth, necessitating PCIe retimers to meet the required connection distances at faster speeds. PCIe is the industry standard for inside-server-system connections between AI accelerators, GPUs, CPUs and other server components. AI models are doubling their computation requirements every six months1 and are now the primary driver of the PCIe roadmap, with PCIe Gen 6 becoming a requirement.
PCIe Gen 6, which operates at 64 gigatransfers-per-second (GT/s), is the first PCIe standard to use four-level pulse-amplitude modulation (PAM4) signaling, displacing the non-return to zero (NRZ) modulation used for the last 20 years. The higher bandwidth and faster data rate limit the physical reach signals can travel reliably, reducing the distance connections can span. Marvell® Alaska P retimers address this by compensating for the signal degradations and regenerating the signal to deliver reliable communication over the physical distances required for connections between GPUs and CPUs within an AI server, between GPUs on different boards, or between CPUs and a pool of shared memory enabled by CXL, among other use cases. The retimers can be used on AI accelerator baseboards, server motherboards, riser cards, or integrated into active electrical cables (PCIe AEC) and active optical cables (PCIe AOC) for emerging multi-rack server system architectures.
“Signal distance is a real dilemma for service providers. We estimate over 75% of cloud and AI servers shipping two years from now will rely on retimers and these servers will contain multiple retimers. An eight GPU server might contain 16 or more of these devices. Retimers will even percolate into enterprise servers,” said Alan Weckel, co-founder of 650 Group. “Marvell’s extensive experience in PAM4 technology positions the company well to drive the evolution of this emerging market.”
Extending PAM4 Leadership
Marvell pioneered PAM4 technology over a decade ago and leads the industry in PAM4 interconnect shipments, enabling data center backend and frontend networks. Today, most of these data center optical interconnects are based on PAM4 technology. The new Alaska P PCIe retimer product line extends the industry-leading Marvell PAM4 connectivity portfolio beyond Ethernet and InfiniBand interconnects into copper and optical PCIe, CXL and proprietary compute fabric links. The product line addresses connections inside AI and general-purpose server systems, expanding the addressable market for Marvell.
Leveraging its PAM4 SerDes technology leadership combined with its data infrastructure IP platform, Marvell has created a state-of-the-art connectivity platform that is enabling leading cloud data center operators to optimize their infrastructure for their unique architectures and workloads. In 2023, Marvell introduced Nova, the industry’s first 1.6T PAM4 DSP. Marvell has also introduced integrated PAM4 DSPs (Perseus) and DSPs optimized for efficiency (Spica Gen2-T) to serve the broadening spectrum of cloud data center link types and use cases. PAM4 technology is also the foundation of the Marvell Alaska A DSP chips optimized for active electrical cable (AEC) applications.
Alaska P PCIe Retimer Benefits and Features
With typical power consumption of 10 watts, the 5nm 16-lane PCIe 6 retimer operates at the lowest power in the industry today. Available in 8- and 16-lane configurations, the Alaska P PCIe devices can compensate for 40dB of channel loss compared to the spec of 32dB at PCIe 6.
The retimers can be used for on-board or cable copper connections or combined with electrical-to-optical components to produce optical PCIe modules, addressing different cloud customer data center architectures. Marvell is working with cable and optical module partners to integrate the products into cloud-optimized interconnect solutions for different data center customer applications.
Key features include:
“Marvell is the industry leader for data center-to-data center, switch-to-switch, rack-to-rack, server-to-switch, and server-to-server connectivity. We are now entering the market for compute fabrics as PCIe and CXL go through an inflection point, migrating from NRZ to PAM4 technology,” said Venu Balasubramonian, vice president of product marketing, Connectivity Business Unit at Marvell. “Marvell is building on more than 10 years of in-house expertise in PAM4 technology and our industry-leading 5nm PAM4 IP portfolio to enable this transition. Our Alaska P PCIe retimer family is an important addition to the comprehensive Marvell accelerated infrastructure portfolio.”
Strong Ecosystem Support
Raghu Nambiar, corporate vice president, Data Center Ecosystems and Solutions, AMD said: “AMD EPYC data center CPUs and AMD Instinct GPUs are essential to delivering the leadership performance needed for the most demanding and business critical applications. Our close collaboration with Marvell on PCIe technologies ensures our customers’ platforms continue to meet the increased connectivity demands of next-generation AI and HPC workloads.”
Dermot O’Driscoll, vice president of product solutions, Infrastructure Line of Business, Arm said: “As AI and ML applications consume more power, there is a critical need for efficient interconnect technology with faster data rates. Marvell’s expanding portfolio will help system designers deliver cloud-optimized Arm Neoverse-powered compute solutions with the latest in PCIe connectivity and marks a new chapter in our long history of collaboration to advance high-performance, efficient computing.”
Osa Mok, chief marketing officer, InnoLight said: “Data center operators need new solutions to support evolving workloads and architectures. For instance, customers are looking for PCIe optical cables for longer reach large-scale AI accelerator interconnect and CXL disaggregation infrastructure where electrical connectivity reaches its limits. InnoLight and Marvell are leveraging our strong optical connectivity expertise and collaboration history to move the optical PCIe ecosystem forward.”
Jim Pappas, director, Technology Initiatives, Intel said: “AI workloads are driving Intel computing platforms to support PCIe 6 with 64 GT/s data rate to optimize performance and utilization. PCIe retimers are critical to help deliver these platforms. PCIe retimer products such as those from Marvell can enable our platforms to meet our customers’ ever-increasing performance requirements of AI and other high-performance compute workloads.”
Vishwas Rao, vice president of product management, Data and Devices, TE said: “Following our successful PCIe 6 DAC cable demo with Marvell in 2023, TE Connectivity applauds Marvell’s PCIe retimer product portfolio launch. The Alaska P PCIe retimer product line together with TE’s advanced cabling solution will provide data center operators a powerful combination for meeting the performance requirements of mission-critical data center applications.”
Availability
The Marvell Alaska P PCIe 16-lane retimer is sampling now to customers and ecosystem partners; the 8-lane product will sample next quarter. For more information, visit here.
About Marvell
To deliver the data infrastructure technology that connects the world, we’re building solutions on the most powerful foundation: our partnerships with our customers. Trusted by the world’s leading technology companies for over 25 years, we move, store, process and secure the world’s data with semiconductor solutions designed for our customers’ current needs and future ambitions. Through a process of deep collaboration and transparency, we’re ultimately changing the way tomorrow’s enterprise, cloud, automotive, and carrier architectures transform—for the better.
Footnote 1: Compute Trends Across Three Eras of Machine Learning
# # #
Marvell and the M logo are trademarks of Marvell or its affiliates. Please visit www.marvell.com for a complete list of Marvell trademarks. Other names and brands may be claimed as the property of others.
AMD, the AMD logo, AMD Instinct, EPYC and combinations thereof are trademarks of Advanced Micro Devices, Inc.
This press release contains forward-looking statements within the meaning of the federal securities laws that involve risks and uncertainties. Forward-looking statements include, without limitation, any statement that may predict, forecast, indicate or imply future events or achievements. Actual events or results may differ materially from those contemplated in this press release. Forward-looking statements are only predictions and are subject to risks, uncertainties and assumptions that are difficult to predict, including those described in the “Risk Factors” section of our Annual Reports on Form 10-K, Quarterly Reports on Form 10-Q and other documents filed by us from time to time with the SEC. Forward-looking statements speak only as of the date they are made. Readers are cautioned not to put undue reliance on forward-looking statements, and no person assumes any obligation to update or revise any such forward-looking statements, whether as a result of new information, future events or otherwise.
For further information, contact:
Michael Kanellos
pr@marvell.com