Intel Xeon Platinum 8592+ Emerald Rapids 64-Core Processor

The Intel Xeon Platinum 8592+ is Intel's flagship Emerald Rapids processor, built on the Intel 7 process node and representing the top of the 5th Generation Intel Xeon Scalable family. It delivers 64 cores and 128 threads on a single socket, with a base frequency of 1.9 GHz and a maximum turbo frequency of 3.9 GHz, backed by 320 MB of Intel Smart Cache (L3). The processor supports up to 12 channels of DDR5-5600 memory per socket, enabling exceptional memory bandwidth for data-intensive enterprise workloads, and accommodates up to 4.5 TB of addressable memory per socket when combining DDR5 and Intel Optane persistent memory configurations.

The 8592+ integrates Intel Advanced Matrix Extensions (AMX) with BFloat16 and INT8 support, providing substantial on-chip inference acceleration without requiring a discrete accelerator for many AI inference deployments. It also incorporates Intel In-Memory Analytics Accelerator (IAA), Intel Data Streaming Accelerator (DSA), Intel QuickAssist Technology (QAT), and Intel Dynamic Load Balancer (DLB) as built-in accelerator engines, reducing CPU overhead for compression, cryptography, and network workload steering. The processor carries a 350 W base TDP, with Thermal Design Power configurable down to lower profiles for thermally constrained environments.

Designed for dual-socket (2S) platforms on the LGA 4677 socket using Intel FCLGA4677 packaging, the 8592+ is broadly deployed in hyperscale and enterprise environments running large-scale virtualization, in-memory databases, AI inference serving, and high-frequency financial analytics. Its PCIe 5.0 and CXL 1.1 lane support enable high-bandwidth connectivity to NVMe storage, GPUs, and memory expansion devices, making it a versatile foundation for modern composable and accelerated infrastructure.

Ideal for

• Large-scale enterprise server virtualization requiring high vCPU density and memory bandwidth per socket
• On-premises AI inference serving leveraging AMX BFloat16 and INT8 acceleration for transformer-based models
• In-memory database and real-time analytics platforms such as SAP HANA and Redis Enterprise
• High-performance cloud-native workloads including containerized microservices and Kubernetes node pools
• Financial services low-latency trading and risk analytics applications demanding high core count and fast memory subsystems
• Data-intensive storage server and software-defined storage deployments utilizing integrated QAT for inline compression and encryption offload

Technical specifications

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