Lenovo 4XG7A63594 SR645 Epyc 7713

Lenovo 4XG7A63594 AMD EPYC 7713 64-Core Processor for ThinkSystem SR645

Overview

The Lenovo 4XG7A63594 is an AMD EPYC 7713 processor option for the ThinkSystem SR645 server platform — a 64-core, 128-thread compute module running a 2 GHz base clock with a 3.675 GHz boost, backed by 256 MB of L3 cache. If you're provisioning SR645 nodes for high-density AI inference workloads, video analytics at scale, or large-scale virtualization, the 7713's core count and memory bandwidth profile make it the CPU to evaluate first in the Milan generation lineup.

Key Features

• 64 Cores / 128 Threads: The EPYC 7713 gives you 128 logical processors per socket — critical for workloads that parallelize aggressively, such as running simultaneous AI analytics streams, containerized microservices, or multi-tenant virtual machines. More threads per socket means fewer servers needed to hit the same aggregate throughput.
• 2 GHz Base / 3.675 GHz Boost: The base clock is conservative by design — EPYC Milan is architected for sustained multi-threaded throughput, not single-threaded burst. The 3.675 GHz boost engages for lightly-threaded tasks, so latency-sensitive operations like database lookups or real-time event processing don't get stranded at base frequency.
• 256 MB L3 Cache: That's a substantial on-die cache footprint — large enough to hold frequently accessed datasets in silicon rather than reaching out to DDR4. For analytics applications or any workload with heavy data reuse, this meaningfully reduces memory latency and pressure on the memory subsystem.
• Socket SP3 — Single or Dual Socket Capable: SP3 supports dual-socket SR645 configurations, letting you scale to 128 cores / 256 threads per 1U or 2U node without any additional switching infrastructure. Plan your rack density accordingly — two 7713s in a single SR645 is a dense compute node.
• DDR4-3200 Memory Support: At 3200 MHz, you get full memory bandwidth from the EPYC platform's eight-channel memory controller. This matters for bandwidth-hungry workloads — the 7713's memory throughput ceiling is well above what most enterprise workloads saturate, which means memory is rarely the constraint.
• 225 W TDP (Configurable to 240 W Up / 225 W Down): The 225 W TDP is the baseline power envelope. The SR645's cooling solution needs to be matched to this — Lenovo does not include a cooler with the 4XG7A63594, so verify your chassis cooling configuration before ordering. The 240 W cTDP-up option can be enabled for additional frequency headroom if thermals allow.
• 64-Bit Operating Mode: Native 64-bit execution means full compatibility with modern server OS environments — RHEL, Windows Server, VMware ESXi, and any enterprise hypervisor. No legacy mode constraints to work around.
• AMD EPYC Milan Architecture: The 7713 is a Milan-generation processor built on TSMC 7nm, giving it competitive performance-per-watt versus prior generations. For server infrastructure refresh cycles, this translates to better density without proportional power budget increases.

Integration and Compatibility

The 4XG7A63594 is a Lenovo-qualified processor option specifically for the ThinkSystem SR645 platform — it ships as a Lenovo-branded component validated against SR645 firmware, BIOS, and thermal specifications. Installing a non-qualified CPU on an SR645 risks compatibility issues at the firmware level, so using Lenovo-sourced options is the correct path for production deployments.

For teams building out rack server deployments around AMD EPYC for AI or video analytics workloads, the SR645 with dual 7713s handles the compute side well. Pair it with appropriate network switching infrastructure and adequate storage I/O to avoid bottlenecks downstream of the CPU. The 7713's 128-thread capacity makes it worth profiling your actual workload thread counts before committing — if your application tops out at 32 threads, the additional core count delivers diminishing returns at this TDP.

If you're deploying SR645 nodes in a cluster for distributed video management or NVR workloads, the 7713's memory bandwidth and core density make it a strong fit for host-based analytics engines that process multiple high-resolution streams concurrently. Match storage and NIC provisioning to the CPU's capabilities — underpowering the I/O stack against this processor wastes the compute headroom you're paying for.

Frequently Asked Questions

Q: Is the Lenovo 4XG7A63594 compatible with the ThinkSystem SR645?

A: Yes. The 4XG7A63594 is a Lenovo-qualified processor option specifically validated for the ThinkSystem SR645 platform, designed to work with SR645 firmware and thermal configurations.

Q: Does the 4XG7A63594 include a cooler?

A: No. A cooling solution is not included with the 4XG7A63594. Verify that your SR645 chassis has the appropriate heatsink and airflow configuration for the 225 W TDP before deployment.

Q: What is the TDP of the AMD EPYC 7713 processor in this kit?

A: The base TDP is 225 W, with a configurable TDP-up of 240 W and a configurable TDP-down of 225 W. Enable cTDP-up only if your chassis cooling supports the higher thermal load.

Q: How many cores and threads does the EPYC 7713 provide?

A: The AMD EPYC 7713 delivers 64 cores and 128 threads per processor. In a dual-socket SR645, that scales to 128 cores and 256 threads per node.

Q: What memory speed does the EPYC 7713 support?

A: The 7713 supports DDR4-SDRAM at up to 3200 MHz, utilizing the EPYC platform's eight-channel memory controller for high aggregate bandwidth.

Q: What is the boost clock speed of the EPYC 7713?

A: The EPYC 7713 boosts to 3.675 GHz from its 2 GHz base clock, engaging higher frequencies for lightly-threaded or latency-sensitive tasks.

Eden Phillips

The 4XG7A63594 is a dense compute option — 64 cores at 2 GHz base with 256 MB of L3 cache is a meaningful profile for any SR645 deployment running parallel workloads. I'd specifically look at this part for AI inference pipelines or multi-stream video analytics where thread count is the primary limiting factor.

Technical Highlights:

• 64 Cores / 128 Threads: At 128 logical CPUs per socket, the 7713 gives analytics and virtualization workloads room to run without contention — useful when you're running dozens of concurrent inference or video decode jobs per host.
• 256 MB L3 Cache: One of the largest on-die cache footprints in the Milan generation — keeps frequently accessed model weights or stream buffers in silicon, measurably reducing latency for data-reuse-heavy workloads.
• 3.675 GHz Boost Clock: The 225 W TDP is managed conservatively at 2 GHz base, but the 3.675 GHz boost means latency-sensitive sequential tasks — real-time event triggers, alarm processing — get frequency headroom when threads aren't fully loaded.

Deployment Considerations:

• No cooler is included — the SR645 chassis heatsink must be configured separately and rated for the 225 W (or 240 W cTDP-up) envelope before the system goes live.
• Socket SP3 dual-socket capability means you can double to 128 cores per node, but verify your SR645 chassis power delivery supports two 225 W CPUs simultaneously — that's 450 W of CPU TDP alone before DRAM, NICs, and drives.

This part is the right call for SR645 nodes provisioned as high-density AI compute or large-scale VMware/KVM hosts where core count and cache depth are the primary sizing criteria — not single-threaded applications where a lower core-count, higher-clock Milan variant would serve better.