Lenovo 4XG7A72959 ST650 V2 Gold 6336Y 24C 185W 2.4GHZ

Lenovo 4XG7A72959 Intel Xeon Gold 6336Y 24-Core Processor Option for ThinkSystem ST650 V2

Overview

The Lenovo 4XG7A72959 is an Intel Xeon Gold 6336Y processor option kit designed for the ThinkSystem ST650 V2 tower server. Packing 24 cores at a 2.4 GHz base clock with a 3.6 GHz turbo, this 3rd Generation Intel Xeon Scalable CPU targets compute-intensive workloads — virtualization hosts, surveillance recording servers, database back-ends, and on-premises analytics platforms where thread count and memory bandwidth matter more than single-core clock speed. If you're sizing a rack or tower server for a mid-to-large deployment and need a processor that balances core density with thermal headroom, the 6336Y is a rational mid-tier pick in the Xeon Gold 6000 series.

Key Features

• 24 Cores / 48 Threads at 2.4 GHz Base: With 48 logical processors available to the OS, the ST650 V2 configured with the 4XG7A72959 can run dozens of concurrent VMs or containerized workloads without contention. For surveillance environments running a VMS, analytics engines, and database services on a single server, this thread count eliminates the bottleneck that 8- or 12-core predecessors introduced under peak recording load.
• 3.6 GHz Turbo Boost Frequency: Burst workloads — video transcoding, live analytics inference, bulk database queries — get a 50% clock uplift over base when thermal conditions allow. This matters in mixed workload scenarios where background services run at base while a foreground process needs sustained single-thread performance.
• Asymmetric Core Architecture (8 High-Priority + 16 Standard Cores): The 6336Y implements a hybrid priority structure: 8 cores run at 2.5 GHz and 16 run at 2.2 GHz under load distribution. Schedulers that are NUMA-aware will direct latency-sensitive threads to the higher-priority cluster — relevant for real-time video decoding or transactional database operations where tail latency is measured.
• 10nm Intel Ice Lake Architecture: Third-generation Xeon Scalable on 10nm delivers improved instructions-per-clock versus the 14nm Cascade Lake generation, plus integrated DL Boost (VNNI instructions) for AI inference acceleration without a discrete GPU — useful if you're running on-camera or server-side analytics workloads that use INT8 quantized models.
• 11.2 GT/s System Bus (UPI) with LGA 4189 Socket: The 11.2 GT/s Ultra Path Interconnect supports dual-socket configurations on compatible ST650 V2 motherboards. If your workload outgrows a single-socket build, a second 4XG7A72959 can be added without a platform change — preserving the server investment while doubling core count to 48 cores / 96 threads.
• 64-Bit Operating Modes: Full x86-64 compatibility means no constraints on OS or hypervisor selection — ESXi, Hyper-V, RHEL, Ubuntu Server, and Windows Server all run natively without emulation overhead.
• 185W TDP: At 185W, this processor sits in the upper range of the Xeon Gold 6000 family's thermal envelope. The ST650 V2's cooling system is rated to handle this TDP, but rack planners should account for the full processor power draw when calculating PDU and UPS capacity — particularly in multi-node deployments.

Integration and Compatibility

The 4XG7A72959 is a Lenovo option kit specific to the ThinkSystem ST650 V2 platform. It ships as a factory-configured or field-installable upgrade through Lenovo's server ecosystem. The LGA 4189 socket is exclusive to the 3rd Generation Xeon Scalable platform — it is not backward-compatible with earlier Xeon server generations that used LGA 3647. Verify your ST650 V2 firmware level before installation, as Lenovo issues UEFI updates to enable support for specific processor stepping revisions.

For surveillance-specific deployments, the 6336Y's integrated DL Boost capability can accelerate inference on compatible network video recorders and VMS platforms that expose CPU-based AI acceleration paths — reducing the need for discrete GPU add-in cards in mid-scale analytics deployments. Pair this server with a purpose-built PoE switch infrastructure and a supported VMS for a validated end-to-end architecture. When planning storage for continuous recording workloads, review the surveillance-grade hard drive options matched to the ST650 V2's drive bays.

This processor option is sourced factory-new with no grey-market or parallel import exposure, consistent with Lenovo's channel-direct supply model for ThinkSystem components.

Frequently Asked Questions

Q: Is the Lenovo 4XG7A72959 compatible with ThinkSystem ST650 V2 servers already in the field?

A: Yes — the 4XG7A72959 is a Lenovo option kit validated for the ThinkSystem ST650 V2 platform. Verify your server's UEFI/firmware level meets Lenovo's minimum requirement for the 6336Y processor stepping before installation.

Q: Can the ST650 V2 support two of these processors in a dual-socket configuration?

A: The ST650 V2 supports dual-socket configurations via the LGA 4189 platform. A second 4XG7A72959 can be installed to double the core count to 48 cores and 96 threads. Check the ST650 V2 configuration guide to confirm your specific chassis and motherboard revision support dual-socket operation.

Q: What is the base and turbo clock speed of the Intel Xeon Gold 6336Y?

A: The Xeon Gold 6336Y runs at a 2.4 GHz base frequency and turbos up to 3.6 GHz. The processor uses an asymmetric core layout: 8 high-priority cores at 2.5 GHz and 16 standard cores at 2.2 GHz under full load.

Q: Is the LGA 4189 socket backward-compatible with older Xeon platforms?

A: No. LGA 4189 is specific to the 3rd Generation Intel Xeon Scalable (Ice Lake) platform and is not compatible with earlier-generation LGA 3647 Xeon servers. This processor cannot be installed in a server that uses an older socket.

Q: What is the TDP of the Xeon Gold 6336Y, and does the ST650 V2 support it?

A: The Xeon Gold 6336Y has a 185W TDP. The ThinkSystem ST650 V2 is designed to support this thermal envelope. Rack planners should account for 185W per processor when calculating PDU and UPS capacity for the installation.

Q: Does this processor support AI or deep learning inference workloads without a GPU?

A: Yes. The 6336Y includes Intel DL Boost (VNNI instructions), which accelerates INT8 inference workloads on-CPU. For VMS platforms and analytics engines that support CPU-based inference, this can reduce or eliminate the need for a discrete GPU add-in card in mid-scale deployments.

Marty Allison

The 4XG7A72959 is the processor option I'd specify for an ST650 V2 build that needs to run a mid-scale VMS alongside on-server analytics without immediately reaching for a GPU. The 6336Y's 24 cores and 48 threads at 2.4 GHz give you enough headroom to separate recording, indexing, and inference workloads into isolated NUMA-pinned VM guests — something 12- or 16-core options can't do cleanly once camera counts climb past 64 channels.

Technical Highlights:

• 24C/48T at 2.4 GHz base, 3.6 GHz turbo: The 50% turbo headroom is what makes this processor work for bursty VMS workloads — base handles sustained recording while turbo absorbs motion-event spikes and export tasks without starving background services.
• Asymmetric 8+16 core layout: The 8 high-priority cores running at 2.5 GHz are your real-time path. Pin latency-sensitive processes (live decode, analytics inference) to these cores and let the 16 standard cores handle archiving and indexing — this is a NUMA/core-affinity configuration step that most VMS vendors don't document but makes a measurable difference at scale.
• Intel DL Boost (VNNI) on 10nm Ice Lake: CPU-based INT8 inference at 10nm IPC levels means you can run lightweight object detection or classification models server-side without a discrete GPU. This is relevant for deployments where a GPU add-in card would push the ST650 V2 over rack power budget or thermal limits.

Deployment Considerations:

• At 185W TDP, this processor will pull meaningful power under sustained load — budget 200–220W per socket including VRM losses when sizing UPS and PDU capacity for the rack.
• LGA 4189 is a dead-end upgrade path in the sense that it doesn't support 4th Gen Sapphire Rapids; if your workload forecast shows 3–5 year growth beyond what 48 threads can handle, evaluate the platform ceiling before committing to this socket generation.

This processor is the right call for a single-site enterprise surveillance server running 64–128 cameras with server-side analytics enabled — it threads the needle between core density, power budget, and platform cost without overbuilding for a deployment that doesn't need GPU-accelerated inference.