The News: Marvell launched the Structera product line of Compute Express Link (CXL) devices built to enable cloud data center operators to address memory performance and scaling challenges in general-purpose servers. Read the full Marvell Technology press release here.
Marvell Unveils Structera CXL Solutions to Meet Hyperscaler Memory Needs
Analyst Take: Marvell unveiled the Marvell Structera product line to directly address the server memory bandwidth and capacity throughout cloud data center environments. The new Structera product line offers innovations tailored for hyperscalers, particularly in the realm of memory expansion and acceleration. The company is addressing critical challenges faced by general-purpose servers in data centers and the cloud, where memory bandwidth has not kept pace with the increasing core count of CPUs. This issue, compounded by limitations in CPU pin count and memory controllers, has hindered the effective utilization of advanced CPUs. Moreover, Marvell is collaborating with key players such as Micron, Samsung, SK Hynix, Intel, AMD, and Arm to ensure broad compatibility and integration of their CXL solutions.
To address memory-intensive applications, data center operators add extra servers to attain higher memory bandwidth and higher memory capacity. The compute capabilities from the added processors are typically not used for these applications, making the servers inefficient, especially from cost and power perspectives. The CXL industry standard addresses this challenge by enabling new architectures that can efficiently add memory to general-purpose servers.
The Promise of CXL Technology and Data Center Trending
CXL technology has been touted as a key technology to address the memory bottleneck issue. While initially anticipated to address much of the AI requirements, we are now seeing the technology focus on addressing in-memory databases and deep learning inference engines such that it would be used as a recommendation engine.
Marvell is addressing another pressing issue in the data center and cloud environments, where there has been a noticeable shift in investment priorities towards AI, leaving traditional server upgrades less prioritized. From our view, hyperscalers are looking for ways to recycle and extend the life of their existing memory resources while upgrading their server infrastructure. This is where Marvell’s focus on CXL technology comes into play, providing a means to reuse server dual in-line memory modules (DIMMs) and optimize memory usage.
Marvell’s New Structera CXL Product Line Breakthroughs
Marvell’s Structera product line comprises two CXL device families that are optimized for different use cases. The Structera A CXL near-memory accelerators are a new category of devices that integrate server-class processor cores and multiple memory channels with CXL to address high-bandwidth memory applications such as deep learning recommendation models (DLRM) and machine learning.
The Structera A 2504 is a CXL 2.0 DDR4 4-channel accelerator that is designed to enable optimal compute and memory scaling including a CXL 2.0/PCIe 5.0 1×16 port controller and up to 200 GB/s memory bandwidth using 4 x DDR5-6400 memory channels and support for up to two DIMMs per channel. The product is underpinned by Marvell 5nm IP and 16 Arm Neoverse V2 (Demeter) cores at 3.2 GHz and supports inline LZ4 compression/decompression, inline AES-XTS 256-bit encryption and decryption as well as embedded hardware security module and secure boot while only requiring a typical power consumption of <100W. This enables the scaling of both compute and memory bandwidth, which is key to enabling new high-performance compute architectures.
As a result, for DLRM applications, Structera A can scale compute and memory bandwidth on a per core basis. For example, take a DLRM server using 64 cores with 400 GB/s of DRAM bandwidth and consume 400W power 6.25 GB/s per CPU core and 1 W per GB/s. Through the implementation of the Structera A CXL accelerator, the DLDM server can now use 80 cores with 600 GB/s of DRAM bandwidth and consume 500W power 7.5 GB/s per CPU core and 0.83W per GB/s. This result shows the ability of a Structera A CXL accelerator to increase the number of cores by 25% and memory bandwidth by 50%.
The Structera X CXL memory-expansion controllers enable terabytes of memory to be added to general-purpose servers and address high-capacity memory applications such as in-memory databases. The Structera CXL device families are the industry’s first to support four memory channels, integrate inline compression, and use 5nm manufacturing processes. Alongside the Structera A and X standard products, Marvell develops custom CXL silicon for cloud operators that is optimized for their unique architectures and workloads.
More Marvell Structera Innovation: Recycling Memory with Structera X 2404 CXL
A key innovation in Marvell’s approach is the use of CXL to recycle memory from older servers, which are typically 5-6 years old. This recycling is facilitated through Marvell’s memory expansion controller, the Structera X 2404, which can handle up to 6TB of DDR4, plus additional gains through compression. Expect this to roll out from the hyperscalers in 2025. The same 2404 will handle DDR5 for future memory recycling.
Specifically, the Stuctera X 2404 is a CXL 2.0 DDR4 4-channel expander that is designed to enable recycling of DDR4 DIMMs to boost server memory capacity including a CXL 2.0/PCIe 5.0 1×16 port or 2×8 controller and up to 100 GB/s memory bandwidth using 4 x DDR5-3200 memory channels and support for up to three DIMMs per channel as well as >6TB of DRAM memory capacity.
Akin to Structera A 2504, the Structera X 2404 product is underpinned by Marvell 5nm IP and supports inline LZ4 compression/decompression, inline AES-XTS 256-bit encryption and decryption as well as embedded hardware security module and secure boot while only needing a typical power consumption of <30W. As such, the solution increases server memory capacity with lower capital expenditure (CapEx) and reduces e-waste.
Structera X 2504: Pioneering Memory Expansion
The Structera X 2504, Marvell’s DDR5 memory expansion controller, is set to be a first mover in the enterprise space. It supports up to 4TB of memory and focuses on in-memory computing and future DDR5 recycling. This controller, built on 5nm technology, incorporates both CXL 2.0 and PCIe 5.0 interfaces, enabling high-speed data transfer and efficient memory utilization.
Key Takeaways: Marvell Structera CXL Delivers Cloud Data Center Server Memory Bandwidth and Capacity Innovation
Overall, we believe the Structera product line launch underscores Marvell’s commitment to addressing the memory bandwidth challenges faced by modern data centers and cloud environments. By leveraging CXL technology for memory expansion and recycling, Marvell aims to provide hyperscalers with the tools needed to optimize their existing infrastructure while meeting the demanding requirements of AI and other data-intensive applications. This strategic focus not only enhances performance but also promotes sustainability by extending the life of existing memory resources.
Disclosure: The Futurum Group is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of The Futurum Group as a whole.
Other Insights from The Futurum Group:
Marvell Q1 Fiscal 2025: Custom AI Silicon Plays the Starring Role
Memcon 2024: Memory Technologies Are a Key Component to Scale AI
Marvell Injects New CXL Technology into Cloud-Optimized CXL Portfolio with Tanzanite Acquisition
Author Information
Camberley brings over 25 years of executive experience leading sales and marketing teams at Fortune 500 firms. Before joining The Futurum Group, she led the Evaluator Group, an information technology analyst firm as Managing Director.
Her career has spanned all elements of sales and marketing including a 360-degree view of addressing challenges and delivering solutions was achieved from crossing the boundary of sales and channel engagement with large enterprise vendors and her own 100-person IT services firm.
Camberley has provided Global 250 startups with go-to-market strategies, creating a new market category “MAID” as Vice President of Marketing at COPAN and led a worldwide marketing team including channels as a VP at VERITAS. At GE Access, a $2B distribution company, she served as VP of a new division and succeeded in growing the company from $14 to $500 million and built a successful 100-person IT services firm. Camberley began her career at IBM in sales and management.
She holds a Bachelor of Science in International Business from California State University – Long Beach and executive certificates from Wellesley and Wharton School of Business.
Ron is an experienced, customer-focused research expert and analyst, with over 20 years of experience in the digital and IT transformation markets, working with businesses to drive consistent revenue and sales growth.
He is a recognized authority at tracking the evolution of and identifying the key disruptive trends within the service enablement ecosystem, including a wide range of topics across software and services, infrastructure, 5G communications, Internet of Things (IoT), Artificial Intelligence (AI), analytics, security, cloud computing, revenue management, and regulatory issues.
Prior to his work with The Futurum Group, Ron worked with GlobalData Technology creating syndicated and custom research across a wide variety of technical fields. His work with Current Analysis focused on the broadband and service provider infrastructure markets.
Ron holds a Master of Arts in Public Policy from University of Nevada — Las Vegas and a Bachelor of Arts in political science/government from William and Mary.
