Skip to content
AI.info

The Pulse

Supermicro Ships H15 Servers With AMD EPYC 9006 Chips

Supermicro has begun shipping its H15 server portfolio with sixth-generation AMD EPYC 9006 processors. The company says the lineup includes more than 20 systems for workloads including AI inference, cloud computing and high-performance computing.

Supermicro Ships H15 Servers With AMD EPYC 9006 Chips

Supermicro is shipping its H15 server portfolio with sixth-generation AMD EPYC 9006 processors for cloud, high-performance computing and agentic AI workloads. The company said in a November 11 release that the lineup includes more than 20 systems, ranging from workstations and edge systems to multi-node and enterprise GPU servers.

The processors were formerly code-named Venice. The H15 portfolio is part of Supermicro's Data Center Building Block Solutions architecture and is optimized for next-generation GPUs, including AMD Instinct accelerators.

Performance claims

Supermicro says the portfolio delivers up to 33% more cores, twice the PCIe bandwidth and 1.6 TB/s of memory bandwidth per socket. For frontier-model host nodes, it claims 1.8 times higher tokens per second than previous generations of AMD CPUs.

The company also claims multiple SPEC CPU world records on its 2U Hyper and CloudDC platforms. Michael McNerney, Supermicro's senior vice president of marketing, said the servers deliver “substantially higher core density, memory bandwidth, and PCIe throughput.”

Systems in the lineup

Hyper is a dual-socket platform designed for enterprise applications, AI inference, virtualization and cloud workloads. CloudDC uses a single socket and is built on the Open Compute Project's Data Center Modular Hardware System specification for cloud-scale environments.

FlexTwin is a liquid-cooled, two-node, dual-CPU system in a 1 OU ORv3 form factor, designed for cloud-native and hyperscale deployments. GrandTwin is a 2U, two-node, single-socket architecture for scale-out environments. Its listed uses include object storage, virtualization, cloud services and high-performance computing.

The 2U Petascale Storage platforms use all-flash storage and support up to 4.8 PB per system. They are designed for uses including software-defined storage-based AI data lakes, large-scale analytics and high-performance computing. The H15 6U SuperBlade supports single- and dual-socket blade configurations, with air- and liquid-cooled versions.

Rack-scale work

Alongside the shipping server portfolio, Supermicro and AMD are collaborating to bring the Helios and Helion rack-scale systems to market. Both are designed for large-scale and agentic AI deployments, but have different areas of focus.

Helios uses AMD Instinct MI455X GPUs and is intended for frontier-model training and high-throughput inference. Helion is built on sixth-generation AMD EPYC CPUs and is designed for compute performance and deployment flexibility. Both are intended to integrate with Supermicro's Data Center Building Block Solutions architecture.

Source

Related stories

Explore

More articles