cancel
Showing results for 
Search instead for 
Did you mean: 

Meet the New AMD EPYC 97X4 Processors, Optimized for Cloud-Native Workloads

Greg_Gibby
Staff
Staff
0 0 5,860

Businesses are increasingly turning to cloud-native applications to innovate and compete successfully in today’s challenging markets. But these are demanding, scalable workloads, requiring modern processors to deliver the most value and efficiency.

Cloud-native applications run in a highly distributed architecture well-suited to running software wherever needed: In public clouds, on-premises enterprise data centers, on the edge, or on Internet of Things (IoT) devices. This flexibility results from software built using microservices running in containers rather than older, monolithic software structures.

Also, cloud-native architectures meet business requirements for rapid application updates. Continuous Integration/Continuous Delivery (CI/CD) deploys application updates in hours rather than the months needed for previous generations of monolithic software.

Increased demand for cloud-native applications challenges cloud services and infrastructure providers to keep up. Costly space and power requirements are particularly problematic as these workloads scale.

The new 4th Gen AMD EPYC 9754 Processor, the highest performing member of the processor family formerly code-named “Bergamo,” can play a significant role in meeting the demands of cloud-native applications. Systems featuring the new processor can provide the robust, energy- and space-efficient infrastructure to run the most demanding and scalable cloud-native services and applications.

Maximize rack density to meet growing demand

AMD EPYC 9754-based systems allow infrastructure operators to maximize rack density and scaling to meet increasing user session demands and deliver new services.

Core density is vital to the 4th Gen AMD EPYC 9754 Processor’s optimization for cloud-native applications. The processor supports up to 128 SMT-capable (Simultaneous Multithreading) energy-efficient processor cores for high performance, 1U and 2U workload density, energy efficiency and X86 compatibility.

High core density means the 4th Gen AMD EPYC 9754 Processor can enable cloud providers to meet customer demand for greater processing power cost-effectively. Systems featuring these CPUs will offer compelling value for space-constrained data centers. The processors feature a 128 cores/256 thread per CPU architecture which will offer the highest virtual Central Processing Unit (vCPU) density and scalability available in a 1U system. A 2-socket system with SMT enabled will provide up to 512 threads per system/server U. That’s extreme density for massive scale of users and services. with many customers preferring to assign each vCPU hosting containers at least one thread, the processors can support more containers per system and rack than other processor options.  This per-system and per-rack scale allows data center operators to accommodate growing demand for high-performance containers, or vCPUs, while minimizing IT infrastructure sprawl and maintaining low latency and high Quality of Service (QoS).

AMD EPYC processor's core density leadership can largely be attributed to its pioneering chiplet-based architecture. AMD was the first processor maker to support chiplets, and we’ve been steadily improving the implementation techniques and capabilities with each generation. The AMD EPYC 9004, which was code-named “Genoa,” introduced in November 2022, scaled to a maximum 96 cores, making it the industry-leading general purpose server processor.SP5-013D Now the EPYC 9754 Processor, specialized for cloud-native applications, goes to new levels of core density and energy efficiency.

Cloud platforms must deliver processing power while maintaining power efficiency. AMD EPYC CPUs power the world’s most energy-efficient servers, enabling new levels of power efficiency while supporting workload scale.EPYC-028C  For example, in third party testing, a 2P AMD EPYC 9754 system delivered ~6.8x the volume of records per second using ~88% fewer Joules for web serving applications compared to 2P Ampere® Altra® Max systems.SP5-141 This same testing showed the 2P AMD EPYC 9754 system scaled No SQL cloud data serving workloads such as Apache Cassandra® with ~2x the op/second writes using ~86% fewer Joules than 2P Ampere Altra Max systems.SP5-142

Broad x86 standardization for EPYC 9004 compatibility

Additionally, EPYC 9754 Processors provide broad x86 compatibility. The processors are socket-compatible with designs for the previously launched AMD EPYC 9004 processor family allowing system manufacturers to develop solutions featuring the new CPUs quickly and easily. Additionally, AMD EPYC 9754 Processors support critical emerging data center architectures such as CXLTM (Compute Express Link), which enable emerging capabilities, such as memory tiering, for additional memory capacity and bandwidth. This type of expansion and adaptability to new composable systems models are vital to assure flexibility and future-readiness that is critical to meet the rapidly-evolving needs of the cloud native data center.

The processor’s x86 compatibility and ecosystems mean systems based on the EPYC 9754 Processor can easily run many current and legacy x86 software and services and modern cloud-native applications. Customers can adopt new container DevOps models and tools in addition to high-density enterprise-class management infrastructure such as VMware®. ARM® architecture does not fully support that broad capability today.4

The EPYC 9754 Processor will deliver value for hyperscale cloud providers. Smaller and more specialized cloud providers will enjoy those benefits as well, as will enterprises increasingly exploring cloud-native software in their on-premises data centers.

Learn more about what EPYC 9754 Processors can do for your business applications. Visit: https://www.amd.com/en/products/cpu/amd-epyc-9754 

 

 

ENDNOTES:

SP5-013D: SPECrate®2017_int_base comparison based on published scores from www.spec.org as of 06/2/2023. Comparison of published 2P AMD EPYC 9654 (1800 SPECrate®2017_int_base, 720 Total TDP W, $23,610 total 1Ku, 192 Total Cores, 2.500 Perf/W, 0.076 Perf/CPU$, http://spec.org/cpu2017/results/res2023q2/cpu2017-20230424-36017.html) is higher than all other 2P competitive processors and is 1.80x the performance of published 2P Intel Xeon Platinum 8490H (1000 SPECrate®2017_int_base, 700 Total TDP W, $34,000 total 1Ku, 120 Total Cores, 1.429 Perf/W, 0.029 Perf/CPU$, http://spec.org/cpu2017/results/res2023q1/cpu2017-20230310-34562.html) [at 1.75x the performance/W] [at 2.59x the performance/CPU$]. Published 2P AMD EPYC 7763 (861 SPECrate®2017_int_base, 560 Total TDP W, $15,780 total 1Ku, 128 Total Cores, 1.538 Perf/W, 0.055 Perf/CPU$, http://spec.org/cpu2017/results/res2021q4/cpu2017-20211121-30148.html) is shown for reference at 0.86x the performance [at 1.08x the performance/W] [at 1.86x the performance/CPU$]. AMD 1Ku pricing and Intel ARK.intel.com specifications and pricing as of 6/1/23. SPEC®, SPEC CPU®, and SPECrate® are registered trademarks of the Standard Performance Evaluation Corporation. See www.spec.org for more information.

EPYC-028C: SPECpower_ssj® 2008, SPECrate®2017_int_energy_base, and SPECrate®2017_fp_energy_base based on results published on SPEC’s website as of 11/10/22. VMmark® server power-performance / server and storage power-performance (PPKW) based results published at https://www.vmware.com/products/vmmark/results3x.1.html?sort=score. The first 80 ranked SPECpower_ssj®2008 publications with the highest overall efficiency overall ssj_ops/W results were all powered by AMD EPYC processors. For SPECrate®2017 Integer (Energy Base), AMD EPYC CPUs power the first 11 top SPECrate®2017_int_energy_base performance/system W scores. For SPECrate®2017 Floating Point (Energy Base), AMD EPYC CPUs power the first 12 SPECrate®2017_fp_energy_base performance/system W scores. For VMmark® server power-performance (PPKW), have the top four results for 2- and 4-socket matched pair results outperforming all other socket results and for VMmark® server and storage power-performance (PPKW), have the top overall score. See https://www.amd.com/en/claims/epyc4#faq-EPYC-028C for the full list. More information about SPEC® is available at http://www.spec.org. SPEC, SPECrate, and SPECpower are registered trademarks of the Standard Performance Evaluation Corporation. VMmark is a registered trademark of VMware in the US or other countries.

SP5-144: Estimated server-side Java® max operations/second comparison based on internal AMD reference platform and Ampere platform measurements for both processors as of 5/2/2023. Testing conducted with 16 vCPUs per group with each group scaled up until the SLA point falls over (typically close to 100% CPU utilization). Configurations: 2P 128-core EPYC 9754 (~393K server-side Java® max operations/second running 16 groups) powered server versus 2P 128-core Altra Max M128-30 (~130K server-side Java® max operations/second running 8 groups) for ~2.6x the performance. OEM published scores will vary based on system configuration and determinism mode used (default performance profile).
SP5-145: Estimated server-side Java® operations/second/W comparison based on internal AMD reference platform measurements for both processors as of 3/31/2023. Configurations: 2P 128-core EPYC 9754 (~29K server-side Java® overall operations/second/W, 720W total TDP) powered server versus 2P 128-core Altra Max M128-30 (~8.4K server-side Java® overall operations/second/W, 500W total TDP) for ~3.5x the performance per system W. OEM published scores will vary based on system configuration and determinism mode used (default performance profile).2 SP5-141: ebizzy records/second comparison based on Phoronix Test Suite paid testing as of 3/31/2023. Configurations: 2P 128-core AMD EPYC 9754 (5,272,631 records/second, 480W avg W, ~868 Joules/run) powered server versus 2P 128-core Altra® Max M128-30 (774,541 records/second, 313.7W avg W, ~7257 Joules/run) for 6.8x the performance using ~88% fewer Joules. Testing not independently verified by AMD. Scores will vary based on system configuration and determinism mode used (320W cTDP performance profile).

3 SP5-142: Apache Cassandra 4.0 stress (writes) comparison based on Phoronix Test Suite paid testing as of 3/31/2023. Configurations: 2P 128- core AMD EPYC 9754 (226,691 op/s, 327.1W avg W, ~3544 Joules/run) powered server versus 2P 128-core Altra® Max M128-30 (114,006 op/s, 174.4W avg W, ~25,981 Joules/run) for 1.98x the performance using 86% fewer Joules. Testing not independently verified by AMD. Scores will vary based on system configuration and determinism mode used (320W cTDP performance profile).

4 Based on VMware Compatibility Guide https://www.vmware.com/resources/compatibility/search.php (as of Apr 19, 2023)