AMD EPYC Processors Scaling to New Heights at SC’24
Supercomputing 2024 has become a stage where technology leaders showcase innovations that push the boundaries of performance, efficiency, and scalability with announcements of the world’s largest supercomputers dominating conversations regarding infrastructure innovation. At this year’s conference in Atlanta, AMD stood tall, solidifying its position as a leading player in the high-performance computing landscape. AMD made a splash as the processor supplier to the world’s new #1 supercomputer, El Capitan, a system that will be run by Lawrence Livermore National Labs clocking in at over a staggering 1.7 Exaflops of performance and tapping a combination of AMD 4th EPYCTM Processors and AMD InstinctTM MI300A APUs. AMD now holds more silicon on the Top 500 than NVIDIA and the #1 and #2 spots on list, showcasing how AMD products are delivering performance value for the scientific community. To build off of this momentum, AMD showcased its new portfolio of silicon products at the show, featuring advancements in system design, the versatile OVR3 rack, and how their work with ecosystem partners is delivering versatile platforms to customers. Let’s unpack the conference highlights.
AMD's Growing Presence in the Top 500
The Top 500 supercomputers list is a badge of honor in HPC, and AMD continues to carve out an impressive space within it. The El Capitan and Frontier systems represent a collective 3 Exaflops of computing performance and over 60% of the top 10 supercomputers’ collective performance. Beyond these top two, several systems powered by AMD EPYC processors and AMD Instinct APUs achieved standout rankings in the Top 500. These aren’t just wins on paper—they represent real-world implications for research institutions and enterprises demanding cutting-edge computational capabilities.
AMD’s secret sauce? Scalability, performance per watt, and a commitment to open standards. With the latest generation of EPYC processors, AMD combines raw, scalable performance with energy efficiency, enabling organizations to hit new heights in scientific simulations, AI model training, and data-intensive analytics. This aligns perfectly with the global demand for sustainable HPC solutions with AMD represented in four of the top ten most energy efficient supercomputers in the Green 500 list.
System Vendors Showcase AMD Platforms
It was a big event, but it seemed that AMD was everywhere! The show floor at SC’24 featured a host of systems featuring AMD’s EPYC processors from across the industry. OEMs and ODMs were quick to highlight AMD’s contributions to their platforms with major players showcasing their AMD-powered servers, underlining how AMD technology enables them to meet evolving workloads.
- ASUS showcased its new servers featuring AMD EPYC 9005 processors and Instinct MI325X APUs aimed squarely at generative AI workloads.
- Dell Technologies showcased EPYC 9005 processors within its liquid-cooled AI factory PowerEdge XE9685L configuration.
- HPE spotlighted its Cray EX systems powering the top-ranked El Capitan supercomputing platform featuring AMD EPYC CPUs and Instinct APUs.
- MSI announced new AMD EPYC 9005 series processor-based servers built on OCP modular hardware system design featuring the latest CXL technology.
- Penguin rolled out its next generation Altus HPC servers with AMD EPYC 9005 series processors targeted for HPC applications.
System vendors underscored the versatility of AMD platforms to address real world use cases, enabling tailored configurations for everything from weather forecasting and precision medicine to generative AI training.
The OVR3 Rack: Modular, Scalable, Future-Ready
A highlight of AMD’s technology on hand at SC’24 was the debut of the OVR3 rack, a new standard rack configuration defined by the Open Compute Project Consortium and quickly being driven into high capacity data center configurations. OVR3 underscores the value of standards-based work that allows for customization for various approaches to network cabling and power delivery, and expands the compute per square foot that can be delivered into data center environments.
The OVR3 rack featured in the AMD booth garnered attention from attendees who noted that the integration of AMD’s EPYC processors and Instinct APUs into a design that prioritizes thermal efficiency and streamlined power distribution hits a critical need for the market. AMD’s commitment to sustainability is evident in the OVR3’s design. By enabling data centers to maximize performance while minimizing energy consumption, AMD is helping companies achieve their compute efficiency goals while delivering performance per rack needed for the AI era.
Key features of the OVR3 include:
- Tool-free design: Simplifies deployment and maintenance.
- Integrated cooling solutions: Optimized for air and liquid cooling to handle high-density configurations.
- Scalability: Designed to grow with organizational needs, whether it’s AI inference, training, or classical HPC.
The configuration also tips AMD’s hat to future rack scale compute delivery with its new acquisition of ZT Systems. Imagine scalable systems delivery of AMD-powered solutions tailored for AI inference and training. The collaboration of a silicon portfolio and highly experienced rack and data center level design is a potent combination that helps AMD further differentiate itself in the market.
Supercomputing Offers a Peak into AMD’s Future
AMD’s technology display at Supercomputing 2024 reflects a broad advancement of innovation, collaboration, and market disruption. As organizations navigate increasingly complex workloads, AMD has built its products as a trusted foundation for the industry, capable of delivering not just performance, but also flexibility and sustainability. The focus on partnerships, like those with leading HPC operators, OEMs/ODMs, and standard configurations like the OCP OVR3 rack, underscores AMD’s holistic approach to meeting market needs. Combined with its growing footprint in the Top 500 and Green 500, AMD is delivering no-compromises solutions to its targets in the HPC market.
As we look ahead, AMD’s silicon portfolio, embrace of standards, and system to rack level innovation will deliver value across data center markets. Whether you’re an enterprise scaling AI workloads, or a research institution tackling the world’s biggest challenges, AMD is proving to be a key ally in the journey.
To learn more, please visit our site.
