For IT managers tasked with upgrading or building data centers, there are many choices to consider when investing in hardware, including whether to select single processor (1P) or dual processor (2P) solutions to fuel their workloads. This document explores how 1P solutions powered by AMD EPYC™ CPUs, can be a cost-effective solution for most workloads. But before we jump into the benefits of 1P platforms, I want to spend a little bit of time talking about why customers are increasingly choosing AMD EPYC™ processors to run their most critical workloads in the data center.
The Rise of AMD EPYC in the Enterprise
AMD EPYC CPUs have become a dominant force in the hyperscale market and are widely deployed across all the major cloud providers. AMD EPYC CPUs are also increasingly being adopted within the enterprise, with many Fortune 500 companies choosing AMD EPYC processor-based infrastructure to run their most critical workloads on-premises. AMD is driving consistent innovation, delivering five generations of EPYC processors since 2017, propelling market share from ~2% to 34%.1 This growth reflects the industry's confidence in the performance, reliability, and total cost of ownership (TCO) advantages that AMD is delivering to customers.
Rethinking the 2P Paradigm
Historically, 2P servers were often the default choice due to perceived requirements for compute power, I/O, memory capacity, and a general "better safe than sorry" mentality. There is also a common belief that 2P platforms provide redundancy – but that is simply not true – if one processor fails, the entire server will fail. Theoretically, 2P servers can actually be less reliable due to the increased number of components and potential points of failure than 1P platforms. Another interesting datapoint based on feedback from our customers is that a significant portion of 2P servers are sold with only one CPU populated – leaving any memory and I/O resources attached to the unpopulated CPU socket unusable.
The 1P Advantage with AMD EPYC
AMD has purpose-designed its 1P processors, so customers aren’t losing out on what they paid for. Moving from two sockets to one does not always translate to giving up on performance or features. Most workloads will run great on 1P AMD EPYC CPU based solutions - providing customers with excellent performance and efficiency without compromise. 1P servers offer several benefits:
- Enhanced Performance: No inter-socket communication bottlenecks.
- Increased Reliability: Fewer components can translate to a lower probability of failure.
- Improved Efficiency: Reduced power consumption and simpler cooling requirements.
- Cost Optimization: Lower acquisition costs and potential software licensing savings.
1P 5th Gen AMD EPYC CPU based solutions deliver impressive performance and scalability, ranging from 8 to 192 cores, up to 128 PCIe® 5.0 lanes, and up to 6TB of memory capacity. For most general-purpose workloads, 1P solutions offer a compelling value proposition.
Choosing the Right Platform: 1P vs. 2P
The following table provides a guideline for selecting between 1P and 2P AMD EPYC solutions:
|
Attribute/ Workload
|
1P
|
2P
|
|
Core Count
|
8-192 (incredible scalability)
|
16-384
|
|
Maximum capacity
|
Up to 6 TB
|
Up to 12 TB
|
|
General purpose workloads
|
Excellent for most general-purpose workloads
|
Best for maximum expandability
|
|
Virtualization workloads
|
Excellent for VM/ container density of ~128 or less
|
Best for maximum VM/ container density
|
|
High Performance Computing (HPC)
|
Excellent for most HPC applications
|
Best for applications that scale with cores and/or memory capacity
|
|
Database analytics workloads
|
Great for small datasets
|
Best for large datasets
|
|
AI workloads
|
Good for inference and edge deployments
|
Best for larger models and training
|
|
Acquisition cost
|
Cost optimized
|
Cost efficient
|
|
Energy efficient
|
Excellent
|
Excellent
|
|
Total Cost of Ownership (TCO)
|
Excellent
|
Excellent
|
Real-World Data Center Modernization
Consider a scenario where a customer is upgrading from 100 x 2P Intel Xeon Gold 6258R (28c) servers. To get the same performance, they can be replaced by 48 x 1P AMD EPYC 9575F (64c) servers and realize:
- ~45% fewer cores and per-core licenses
- Up to 52% less power consumption
- Up to 40% 3-year TCO reduction2
Comparing the latest Intel and AMD offerings, it would take 71 x 1P Intel Xeon Platinum 8592 (64c) servers to deliver the same performance as 48 x 1P AMD EPYC 9575F (64c) servers – with AMD delivering the following advantages:
- ~32% fewer cores and per-core licenses
- Up to 27% less power consumption
- Up to 32% 3-year TCO reduction2
- Up to 48% more performance per server
A robust portfolio of 1P solutions available from leading OEMs
1P AMD EPYC CPU based platforms are available from leading OEMs, including Dell Technologies, Hewlett Packard Enterprise, Lenovo, and Supermicro, among others.
Contact your preferred provider and “Ask for AMD EPYC CPUs”
[1] Source: Mercury Research Sell-in Revenue Shipment Estimates 1H24
[2] 9xxTCO-007: This scenario contains many assumptions and estimates and, while based on AMD internal research and best approximations, should be considered an example for information purposes only, and not used as a basis for decision making over actual testing. The AMD Server & Greenhouse Gas Emissions TCO (total cost of ownership) Estimator Tool - version 1.3, compares the selected AMD EPYC™ and Intel® Xeon® CPU based server solutions required to deliver a TOTAL_PERFORMANCE of 39700 units of SPECrate®2017_int_base performance as of January 6, 2025. This estimation compares upgrading from a legacy 2P Intel Xeon 28 core Gold_6258R based server with a SPECrate2017_int_base score of 397, https://spec.org/cpu2017/results/res2020q3/cpu2017-20200915-23981.pdf) versus 1P AMD 64 core EPYC_9575F powered server with a SPECrate2017_int_base score of 833, https://spec.org/cpu2017/results/res2024q4/cpu2017-20241104-45211.pdf ; compared to upgrading to a 1P Intel Xeon 64 core Platinum_8592+ based server with a SPECrate2017_int_base score of 563, https://spec.org/cpu2017/results/res2024q3/cpu2017-20240729-44257.pdf;
Environmental impact estimates made leveraging this data, using the Country / Region specific electricity factors from Country Specific Electricity Factors - 2024, and the United States Environmental Protection Agency Greenhouse Gas Equivalencies Calculator. For additional details, see https://www.amd.com/en/claims/epyc.html#q=9xx5TCO-007.
