#United States #Santa Clara #AMD EPYC #Intel Xeon #Marvell Tech

Marvell Technology Achieves Interoperability Success

Marvell Technology, Inc., a prominent player in semiconductor solutions for data infrastructure, has announced a notable achievement in the tech world. Their Structera portfolio—comprised of Compute Express Link (CXL) devices—has successfully integrated with AMD EPYC CPUs and 5th Gen Intel Xeon Scalable platforms. This milestone underscores Marvell's dedication to fostering an interoperable CXL ecosystem, which is crucial in meeting the escalating demands for memory bandwidth and capacity in future cloud data centers.

A Collaborative Effort

In collaboration with industry giants AMD and Intel, Marvell conducted comprehensive interoperability tests across diverse configurations, workloads, and operating environments. The results revealed a seamless integration process, offering the stability and scalability that cloud data center operators are keenly seeking for large-scale deployments. This achievement empowers customers with the flexibility to select from various CPU architectures, facilitating deployment across different hardware configurations. As a result, users are able to leverage the unique strengths of each platform, optimizing investments while ensuring smooth integration and consistent performance.

Will Chu, Sr. Vice President and General Manager at Marvell, commented, "CXL 2.0 is paving the way for scalable infrastructure growth. Our collaboration with AMD and Intel is instrumental in nurturing an ecosystem that fully realizes the benefits of CXL for clients around the globe. This interoperability highlights our capability to deliver cutting-edge, high-performance solutions that enhance the efficiency and effectiveness of cloud technology."

Powering High-Performance Data Centers

With the advancements of Structera, cloud service providers and server OEMs can significantly increase server memory capacity and bandwidth. Not only can they augment computing cores to boost performance for memory-intensive applications—such as artificial intelligence, machine learning, and high-performance computing—but they can do this in a cost-effective manner. The Structera product line features key innovations such as the Structera A and Structera X:

• - Structera A: This near-memory accelerator boasts server-class Arm Neoverse V2 cores and DDR5 support, allowing for up to 1.6 Tbps of memory bandwidth, optimized specifically for high-bandwidth workloads like deep learning recommendation models.
• - Structera X: These memory expansion controllers back both DDR4 and DDR5 DIMMs, facilitating terabyte-level memory deployments, all while emphasizing cost efficiency and power consumption.

By leveraging CXL 2.0 technology, these products enable cache coherency and memory pooling across multiple devices, optimizing resource utilization. Consequently, this reduces the necessity for additional servers, promoting sustainable data center architectures—a significant advantage in today’s market.

Marvell's Vision and Commitment

Marvell is on a mission to transform the way data infrastructure technology connects the global landscape. With over 25 years of experience, they are trusted by leading tech firms around the world to facilitate data movement, storage, processing, and security. Through collaborative partnerships and a commitment to innovation, Marvell’s vision is centered on efficiently meeting the current demands and future expectations of enterprise, cloud, automotive, and carrier architectures.

In conclusion, the interoperability achievement between Marvell's CXL portfolio and processors from AMD and Intel represents a significant stepping stone toward the future of scalable cloud data center technology. This collaboration not only demonstrates the potential of CXL technology but also presents new opportunities for improved memory efficiency and cost management in cloud infrastructures.