Data centers are the backbone of the modern digital economy, powering a wide range of applications and services that require high performance, energy efficiency, and ease of adoption. Whether you are running online shopping, web hosting, data analysis, video streaming, or artificial intelligence (AI) workloads, you need a processor that can deliver the best results without compromising on any of these aspects, all with seamless portability so that you can focus on your primary business.
The growing importance of data center business and the vast amounts of power now dedicated to data center and cloud infrastructure has created motivation for a number of vendors to propose alternative processor choices often promising advantages over common x86 solutions. One of these new alternatives is from Nvidia with their “Grace Superchip” based on ARM processor IP. Typically, these are introduced with great fanfare and claims of significant performance and energy efficiency benefits relative to x86. Far too often, these claims are quite difficult to translate into real world competitive workload scenarios—with out-of-date, sub-optimized alternatives or poorly documented assumptions.
AMD EPYC processors continue to set new standards in data center performance, power efficiency, security, and total cost of ownership, driven by continuous innovation on the proven x86 architecture. Whether deployed on-premise, in cloud environments or across diverse industries, the 4th Gen AMD EPYC processor portfolio offers cutting-edge solutions to address varied workload requirements. The extensive AMD EPYC ecosystem comprises over 250 distinct server designs and supports nearly 900 unique cloud instances and is trusted by some of the largest companies in the world to run their services. AMD EPYC processors hold over 300 world records for performance and efficiency across a broad spectrum of benchmarks, including business applications, technical computing, data management, data analytics, digital services, media and entertainment, and infrastructure solutions.
EPYC leads against ARM based solutions in numerous industry standard benchmarks as we show in this article. Furthermore, with the tried-and-tested x86-64 architecture pioneered by AMD, you can get this without expensive porting or architectural transitions.
Performance Comparison
We compared the performance of AMD EPYC processors and NVIDIA Grace CPU Superchip across ten key workloads, covering general purpose computing, server-side Java, power efficiency, transactional databases, decision support systems, web servers, in-memory analytics, video encoding, and HPC workloads. We used industry-standard benchmarks and testing methodologies to ensure fair and transparent comparisons. A complete set of test results with discussions of the testing and documentation of system and test configurations is published in a blog by Raghu Nambiar at this link. A summary of the results are shown in Figure 1 as a ratio of the performance of AMD EPYC processors relative to the performance of NVIDIA Grace CPU Superchip systems.
Figure 1: Performance comparison of AMD EPYC processors and NVIDIA Grace CPU Superchip
As you can see, AMD EPYC processors deliver more than twice the performance of the NVIDIA Grace CPU Superchip across workloads representative of multiple verticals, demonstrating the superior capabilities of AMD EPYC processors for data center performance without compromises.
Efficiency Leadership
It is important to note that the AMD EPYC architecture gives you performance and energy efficiency out of the box. You get all the performance you need for critical apps while also helping to meet energy goals without having to change architectures or even system suppliers. Leadership energy efficiency is available in the form of AMD EPYC-based servers from the server vendors you know and trust to run your business today. Figure 2 below provides a breakout of the data in the graphic above to focus more directly on the results of systems tested with the well-regarded SPECpower_ssj2008 benchmark designed to showcase energy efficiency in workload processing.
Figure 2: SPECpower_ssj® 2008 relative performance
As shown, single- and dual-socket AMD EPYC 9754-based systems outperform an NVIDIA Grace system by ~2.50x and ~2.75x, respectively. Further, a dual-socket AMD EPYC 9654 system outperformed the same NVIDIA system by ~2.27x on the same tests.[1][2][3]
Compatibility Advantage
Besides performance and efficiency, another important factor for datacenter operators is compatibility. Globally, it is estimated that there are trillions of lines of software code, much of this written for x86 architecture. AMD EPYC processors are based on the x86-64 architecture pioneered by AMD and this architecture is the most widely used and supported architecture in the datacenter industry. This means that you can run a broad set of workloads on AMD EPYC processors without any compatibility issues, and without expensive architectural transitions to a different ISA. It can be easy to forget that a software port is one thing, the real burden comes with having to manage and maintain multiple code bases. Moreover, each ARM implementation varies by the chip vendor, so using one ARM chip does not mean you can expect seamless compatibility with an ARM chip from another vendor.
AMD believes that architectural change is difficult, expensive and fraught with risk. An open standards-based approach balanced with relentless innovation provides customers with a better path. AMD EPYC processors have helped server vendors and ecosystem IHVs support the transition to the latest features and standards, such as PCIe® 5.0, DDR5, and CXL, to ensure future-proofing and interoperability and a smooth adoption path for customers. Few ARM offerings have proven track record in bringing the extended IHV server ecosystem along in co-innovation. AMD continues to execute to a steady roadmap of processor innovation and provides a trusted choice for customers to advance their data center.
Conclusion
AMD EPYC processors are the best choice for datacenter performance and efficiency, as they outperform the NVIDIA Grace CPU Superchip across ten key workloads, based on extensive industry-standard benchmark publications and testing. AMD EPYC processors also offer the advantage of x86 processor architecture compatibility, which enables you to deploy a broad set of workloads with no compromises, and without expensive architectural transitions to a different ISA. AMD EPYC processors are the best option for datacenter operators who want to maximize performance while minimizing power and real estate footprint in a datacenter with an easy button. In the age of AI where you need capacity for your emerging AI workloads, AMD provides the best option with no compromises based on industry standards, transparency in data and benchmarks, and broad availability of platforms and solutions across the Ecosystem without expensive architectural transitions. I encourage you to read Raghu Nambiar’s full performance analysis blog for much more detail and description of methodologies and the full range of workloads covered. And you are welcome to learn more about AMD EPYC technologies and solutions at the EPYC home page.
Endnotes
- SP5-279: As of 07/12/2024, a 1P AMD EPYC™ 9754 system delivers a 2.50x SPECpower_ssj® 2008 overall ssj_ops/watt uplift versus a 2P NVIDIA Grace™ CPU Superchip system. Configurations: 1P 128-core EPYC 9754 (33,014 overall ssj_ops/watt, 2U, https://www.spec.org/power_ssj2008/results/res2023q3/power_ssj2008-20230524-01270.html) versus 2P 72-core Nvidia Grace Superchip (13,218 overall ssj_ops/watt, 2U, https://www.spec.org/power_ssj2008/results/res2024q3/power_ssj2008-20240515-01413.html. SPEC® and SPECpower_ssj® 2008 are registered trademarks of the Standard Performance Evaluation Corporation. See www.spec.org for more information.
- SP5-280: As of 07/12/2024, a 2P AMD EPYC™ 9754 system delivers a 2.75x SPECpower_ssj® 2008 overall ssj_ops/watt uplift versus a 2P NVIDIA Grace™ CPU Superchip system. Configurations: 2P 128-core EPYC 9754 (36,398 overall ssj_ops/watt, 2U, https://www.spec.org/power_ssj2008/results/res2024q2/power_ssj2008-20240327-01386.html) versus 72-core Nvidia Grace Superchip (13,218 overall ssj_ops/watt, 2U, https://www.spec.org/power_ssj2008/results/res2024q3/power_ssj2008-20240515-01413.html. SPEC® and SPECpower_ssj® 2008 are registered trademarks of the Standard Performance Evaluation Corporation. See www.spec.org for more information.
- SP5-278: As of 07/12/2024, a 2P AMD EPYC™ 9654 system delivers a 2.27x SPECpower_ssj® 2008 overall ssj_ops/watt uplift versus a 2P NVIDIA Grace™ CPU Superchip system. Configurations: 2P 96-core AMD EPYC 9654 (30,064 overall ssj_ops/w, 2U, https://www.spec.org/power_ssj2008/results/res2022q4/power_ssj2008-20221204-01203.html) versus 2P 72-core Nvidia Grace Superchip (13,218 overall ssj_ops/watt, 2U, https://www.spec.org/power_ssj2008/results/res2024q3/power_ssj2008-20240515-01413.html. SPEC® and SPECpower_ssj® 2008 are registered trademarks of the Standard Performance Evaluation Corporation. See www.spec.org for more information.