Intel Xeon 6787P (Granite Rapids) 86-Core Processor

The Intel Xeon 6787P is a high-core-count processor from the Granite Rapids generation, built on Intel's Intel 3 process node and designed for demanding dual-socket server deployments. Featuring 86 physical cores with support for Intel Hyper-Threading Technology, the 6787P delivers substantial parallel compute throughput suited to workloads that require both high core density and strong per-core performance. The processor incorporates Intel's updated P-core microarchitecture, offering improved instructions-per-clock efficiency over prior Xeon Scalable generations, along with a substantially enlarged last-level cache to reduce memory latency bottlenecks in data-intensive applications.

The 6787P supports DDR5 memory across eight memory channels per socket, enabling high aggregate memory bandwidth essential for in-memory databases, analytics engines, and large-scale virtualization environments. PCIe 5.0 connectivity provides high-throughput lanes for NVMe storage arrays, high-speed networking adapters, and GPU or accelerator co-processors. The platform also integrates Intel's Advanced Matrix Extensions (AMX) for accelerated AI inference and training workloads directly on the CPU, as well as Intel Data Streaming Accelerator (DSA) and other built-in accelerator engines that offload common data-movement and compression tasks from core execution units.

As a mainstream Granite Rapids SKU, the 6787P is widely specified in two-socket rack servers targeting enterprise consolidation, cloud-adjacent infrastructure, and technical computing. Its combination of 86 cores, broad accelerator integration, and DDR5/PCIe 5.0 platform support positions it as a versatile foundation for workloads ranging from large-scale virtual machine hosting and containerized microservices to EDA simulation, financial risk modeling, and real-time AI inference pipelines.

Ideal for

• Large-scale server virtualization and VDI consolidation, leveraging 86 cores per socket to host high VM densities with strong per-VM performance isolation
• In-memory and real-time analytics databases such as SAP HANA and Oracle In-Memory, benefiting from DDR5 memory bandwidth and expanded last-level cache
• Enterprise AI inference and batch inferencing pipelines utilizing Intel AMX instructions to run quantized deep learning models without discrete accelerators
• High-throughput containerized microservices and Kubernetes clusters where per-node core count directly determines pod density and scheduling efficiency
• Electronic design automation (EDA) and computational simulation workloads requiring many parallel single-threaded jobs running simultaneously across a large core pool
• Financial services risk calculation and Monte Carlo simulation engines that benefit from high core parallelism, large cache capacity, and low-latency DDR5 memory access

Technical specifications

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