AMD EPYC 9475F server CPU

Experience out-of-the-box performance and density leadership for the growing demands of AI-enabled, business-critical data center workloads. More from the Manufacturer

AMD EPYC Model 9475F

Experience out-of-the-box performance and density leadership for the growing demands of AI-enabled, business-critical data center workloads

Cloud and Hyperscale deployments and Enterprise IT

AI and AI-enabled workloads

AMD EPYC™ 9005 Processors are the newest generation of the powerful and efficient AMD EPYC processor family for servers that have set more than five hundred performance and efficiency world

records

. Advancements in the EPYC 9005 processor family are enabled by the breakthrough high performance, highly efficient "Zen5” processor core architecture and advanced microprocessor process technologies to better meet the needs of the modern AI-enabled data center. The complete line of processor offerings include a wide range of core counts (up to 192 cores, 384 threads per CPU), frequencies (up to 5 GHz), cache capacities, TDP levels and competitive cost points-- all complemented by the familiar x86 software compatibility that allow AMD EPYC 9005-based servers to readily support almost any business need.

In many ways, the AMD EPYC 9005 can be considered the world’s best data center processor—and one that is readily accessible and easily deployable. From supporting corporate AI-enablement initiatives to powering large-scale cloud-based infrastructures to hosting the most demanding business-critical applications, AMD EPYC 9005-based servers offer leadership performance, density and efficiency to meet customer needs with minimal software rework.

Core-for-core, the new EPYC 9005 processor brings higher performance than even the leadership performance of the AMD EPYC 9004 family—with the AMD EPYC 9005 offering up to a geomean 37% instruction-per-clock (IPC) single-thread uplift compared to AMD EPYC 9004 processors for machine learning and high performance computing workloads, and geomean 17% IPC single-thread uplift compared to AMD EPYC 9004 processors for general purpose enterprise computing workloads

(9xx5-001)

. In the past seven years AMD has relentlessly driven server processor performance forward, delivering 5 generations of EPYC processors that have increased core counts by 6X versus the initial top-of-stack EPYC product and driven performance up by 10.9% across those 5 generations.

(9xx5-069A)

As an example of the outstanding general-purpose performance of the new platform, consider that the AMD EPYC 9965 dual-CPU system outperforms the highest core count dual Intel® Xeon® Platinum 8592+ system by ~2.65X based on published SPECrate®2017_int_base performance metrics

(9xx5-002D)

Systems based on the AMD EPYC 9005 processor will support IT initiatives from data center consolidation and modernization to increasingly demanding enterprise application needs. Customers replacing 1000 aging dual Intel® Xeon® 8280 “Cascade Lake” systems can find up to 7:1 workload consolidation opportunities by moving the work of the older systems to 131 dual processor AMD EPYC 9965 based servers-- freeing up space and power in the data center to support new innovation

(SP9xx5TCO-002)

. AMD EPYC processor-based systems demonstrate strong performance and efficiency metrics for many general-purpose enterprise computing workloads: databases, video, and Java®/cloud native workloads, for advanced technical workloads such as CAE, CFD, and FAE, for high performance computing workloads such as modelling and simulation.

AMD EPYC CPU-based servers will be compelling platforms to accommodate expanding AI enablement within the enterprise but also support business imperatives to improve energy efficiency and to reign in data center sprawl through density enabled by virtualization and cloud technologies.

The proven performance, efficiency and easy x86 software compatibility of the AMD EPYC architecture have prompted companies, governments and organizations around the globe to make the switch to AMD EPYC processor-based servers for their most demanding computing tasks. With the leadership AMD EPYC portfolio available in servers from all leading systems vendors and qualified with leading software packages and AI frameworks, models and tools, servers based on the EPYC 9005 family of processors represent a seamless evolutionary path for compute and AI leadership business solutions.

AMD EPYC™ 9005 Processor-based Systems Example Use Cases / Deployments

Cloud and Hyperscale

Deployed to enable high-performance, high-density compute solutions to support the most demanding customer workloads while mitigating needs to grow data center real estate and power needs. Higher core count-based server systems allow cloud providers to offer increasingly large instance configurations—attracting new workloads and revenue.

Enterprise IT Applications

With customers facing demands from alternative platforms to adopt new instruction sets, recompile and maintain new binaries or to do software work to enable new features and accelerators on their latest generation systems, customers may find it easier to switch and migrate workloads hosted on older, less-efficient systems to the consistent AMD x86 architecture.

Customers can deploy AMD EPYC 9005-based servers seamlessly alongside existing x86 server infrastructure to efficiently host new or growing applications.

Customers will likely often target AMD EPYC 9005-based servers for the most demanding workloads in their environments to gain incremental performance, scale and efficiency.

AI and AI-enabled workloads

Customers are interested in the high performance and scalability of AMD EPYC CPU-based systems as excellent infrastructure for workload consolidation initiatives as a first step to help free up real estate, energy and funding to allow for AI technology-driven innovation.

A growing set of AI workloads—Language model (~13 billion parameters and below) inference, classical machine learning workloads, recommendation systems and the ever-expanding set of “traditional” collaboration, database, design and other applications that are AI-enabled—are well suited to running on powerful CPUs without the need for expensive and energy-hungry GPU acceleration.

Large LLMs, real-time AI training and other demanding generative AI workloads typically require one or more advanced GPU solutions. Select high frequency AMD EPYC 9005 processors are optimized for use as host CPUs for GPU-enabled systems, providing the compelling high performance and energy efficiency on select workloads needed to get the most out of such sizeable investments in GPU solutions.

Enterprise Distributed IT (Branch and edge)

Deployed to branch office or departmental/site server where growing workload demands require more performance and increased considerations for energy efficiency.

Deployed into environments where customers seek to deploy the very latest generation of systems with the expectation that they will be deployed for a significantly long lifecycle to mitigate the need to replace or upgrade systems more frequently.