AMD EPYC 9845 (Turin Dense, 160-core Zen 5c)

The AMD EPYC 9845 is a member of the 'Turin Dense' family within the 9005 series, introduced in October 2024, and represents AMD's highest core-count offering in the dense Zen 5c core configuration. With 160 physical cores and 320 threads on a single socket, this processor is purpose-built for environments where thread density and energy efficiency take precedence over single-threaded peak frequency. The Zen 5c architecture employs a compact die layout that packs significantly more cores per unit of silicon area compared to full-fat Zen 5 cores, enabling exceptional compute density without sacrificing the modern IPC improvements inherent to the Zen 5 microarchitecture, including wider front-end dispatch, improved branch prediction, and enhanced AVX-512 execution pipelines.

The EPYC 9845 is fabricated on TSMC's advanced 4nm-class process node, contributing to its favorable power profile relative to its core count. It is housed in the SP5 (LGA-6096) socket and supports eight-channel DDR5 memory per socket with a maximum memory bandwidth exceeding 460 GB/s, ensuring that even highly parallel workloads are not memory-starved. With support for up to 6TB of DDR5 registered ECC memory per socket and compatibility with CXL 1.1+ memory expansion, the EPYC 9845 is positioned for hyperscale data center deployments where both core density and memory capacity matter.

The processor integrates 128 lanes of PCIe 5.0, enabling high-speed connectivity for NVMe storage arrays, high-performance networking adapters, and GPU accelerators. AMD Infinity Guard security features—including SEV-SNP (Secure Encrypted Virtualization with Secure Nested Paging), memory encryption, and hardware root-of-trust—make the EPYC 9845 a compelling choice for confidential computing workloads and multi-tenant cloud infrastructure. Its combination of massive parallelism, modern security primitives, and efficient power utilization positions it as a leading processor for scale-out container orchestration, microservices platforms, and cloud-native application hosting.

Ideal for

• Large-scale Kubernetes and container orchestration platforms where maximizing pod density per physical node reduces infrastructure footprint and operational overhead
• High-density cloud hosting and virtual machine consolidation for public and private cloud providers seeking to maximize vCPU count per rack unit
• Microservices architectures and API gateway infrastructure requiring thousands of concurrent lightweight threads with consistent low-latency response
• Big data analytics pipelines using Apache Spark or Flink, where massive thread parallelism accelerates distributed computation across large datasets in-memory
• Confidential computing and secure multi-tenant environments leveraging AMD SEV-SNP for hardware-enforced VM isolation in shared infrastructure
• Software-defined networking (SDN) and network function virtualization (NFV) platforms that benefit from extreme core counts to run numerous virtual network functions simultaneously

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.