Proven Performance and Energy Efficiency

Achieve exceptional price performance and power savings with proven 3rd Gen AMD EPYC™ processors.1

READ DATASHEET

 

AMD EPYC Logo Wordmark

Update your data center with cost effective, energy efficient solutions.

AMD EPYC™ 7003 Series Processors have set a standard for performance and efficiency for mainstream data center servers. Customers that have done extensive application qualification on DDR4/PCIe®4 generation systems and seek performant, cost-effective solutions to data center challenges, can find tremendous value with AMD EPYC™ 7003 processor-based servers. 

Better Performance per CPU Price2

(64 Total Cores)

2X EPYC 7543 32c
+36%
2X Xeon 8358 32c

Better Energy Efficiency3

AMD EPYC 7763
+85%
Intel Xeon 8380

Cutting-Edge Security Features

With security threats on the rise, you need to ensure maximum protection for your critical data. EPYC processors come with AMD Infinity Guard—a full suite of cutting-edge security features, built into the silicon and designed to defend against internal and external threats.4

AMD Infinity Guard helps decrease potential attack surfaces as software boots, executes, and processes your data. It includes:

  • Secure Encrypted Virtualization (SEV) to help safeguard VM privacy and integrity
  • Secure Nested Paging (SEV-SNP) for strong memory integrity protection capabilities
  • Secure Memory Encryption (SME) to help guard against attacks on main memory
  • AMD Shadow Stack™ for hardware-enforced stack protection capabilities against malware attacks

Energy Efficiency

EPYC processors power the most energy efficient x86 servers, delivering exceptional performance and reducing energy costs.5 EPYC CPUs can help minimize environmental impacts from data center operations while advancing your company’s sustainability objectives.

AMD has even bigger plans in place for the future. Our goal is to deliver 30x increase in energy efficiency for AMD processors and accelerators powering servers for AI-training and HPC from 2020-2025. Our goal equates to a 97% reduction in energy use per computation by 2025. If all AI and HPC server nodes globally were to make similar gains, billions of kilowatt-hours of electricity could be saved in 2025 relative to baseline trends.

Energy Efficiency
Cut total cost of ownership by an estimated 68%

Outstanding Return on IT Investment

Capture the value of your IT investment. Cut total cost of ownership in virtualized environments by an estimated 68% to deliver 320 VMs.6  EPYC processor-powered servers can improve time-to-value for your applications and help you gain business-critical insights faster.

EPYC processor-powered single-socket servers also deliver compute power that is right-sized for your workloads, so that you may be able to satisfy your business requirements without having to scale up to dual-socket servers— thus helping minimize licensing costs and reducing power consumption.

 

Amazon Web Services
Cisco
Dell Technologies
Google Cloud logo
Hewlett Packard Enterprise
Lenovo
Microsoft Azure logo
Supermicro logo

 

Backed by Industry Leaders

Major infrastructure and software providers work with AMD to help ensure your applications work exceptionally well with EPYC processor-powered servers. EPYC processors run virtually all x86 applications, enable worry free migration, and seamlessly integrate into existing x86 infrastructures.

Whether you are seeking virtualization, containerization, hybrid cloud, or software-defined infrastructure opportunities, there is an EPYC processor-powered solution to meet your needs—all thanks to our strong partnerships with these industry leaders.

The Secret is Under the Hood

Built on AMD ‘Zen 3’ microarchitecture-based cores and AMD Infinity Architecture, AMD EPYC 7003 Series processors provide a full feature set across the stack with PCIe Gen4 I/O, 7nm x86 CPU technology, and an integrated security processor on die.

They also deliver up to 32MB of L3 cache per core with standard processors and up to 768 MBs per socket with 3D V-cache™ technology, 4-6-8 memory channel interleaving to optimize performance in multiple DIMM configurations, and synchronized clocks between fabric and memory—technologies that come together to drive leadership performance.

AMD EPYC 7003 Series Processors with AMD 3D V-Cache™ Technology

With the addition of AMD 3D V-Cache™ technology, you can:

  • Help lower TCO while accelerating product development
  • Support sustainability through exceptional energy efficiency
  • Provide the confidence of modern security
  • Socket compatible with existing AMD EPYC 7003 platforms

Model Specifications

AMD EPYC 7003 Series processors scale from 8 to 64 cores (or 16 to 128 threads per socket). No other server processors enable this level of core density. No matter what your CPU performance and throughput needs may be, there is an EPYC processor model to match.

HELP ME CHOOSE

Model # of CPU Cores # of Threads Max. Boost Clock Base Clock L3 Cache Default TDP
AMD EPYC™ 7763 64 128 Up to 3.5GHz 2.45GHz 256MB 280W
AMD EPYC™ 7713P 64 128 Up to 3.675GHz 2.0GHz 256MB 225W
AMD EPYC™ 7713 64 128 Up to 3.675GHz 2.0GHz 256MB 225W
AMD EPYC™ 7663P 56 112 Up to 3.5GHz 2.0GHz 256MB 240W
AMD EPYC™ 7663 56 112 Up to 3.5GHz 2.0GHz 256MB 240W
AMD EPYC™ 7643P 48 96 Up to 3.6GHz 2.3GHz 256MB 225W
AMD EPYC™ 7773X 64 128 Up to 3.5GHz 2.2GHz 768MB 280W
AMD EPYC™ 7643 48 96 Up to 3.6GHz 2.3GHz 256MB 225W
AMD EPYC™ 7573X 32 64 Up to 3.6GHz 2.8GHz 768MB 280W
AMD EPYC™ 75F3 32 64 Up to 4.0GHz 2.95GHz 256MB 280W
AMD EPYC™ 7543P 32 64 Up to 3.7GHz 2.8GHz 256MB 225W
AMD EPYC™ 7543 32 64 Up to 3.7GHz 2.8GHz 256MB 225W
AMD EPYC™ 7513 32 64 Up to 3.65GHz 2.6GHz 128MB 200W
AMD EPYC™ 7473X 24 48 Up to 3.7GHz 2.8GHz 768MB 240W
AMD EPYC™ 7453 28 56 Up to 3.45GHz 2.75GHz 64MB 225W
AMD EPYC™ 74F3 24 48 Up to 4.0GHz 3.2GHz 256MB 240W
AMD EPYC™ 7443P 24 48 Up to 4.0GHz 2.85GHz 128MB 200W
AMD EPYC™ 7443 24 48 Up to 4.0GHz 2.85GHz 128MB 200W
AMD EPYC™ 7413 24 48 Up to 3.6GHz 2.65GHz 128MB 180W
AMD EPYC™ 73F3 16 32 Up to 4.0GHz 3.5GHz 256MB 240W
AMD EPYC™ 7373X 16 32 Up to 3.8GHz 3.05GHz 768MB 240W
AMD EPYC™ 7343 16 32 Up to 3.9GHz 3.2GHz 128MB 190W
AMD EPYC™ 7313P 16 32 Up to 3.7GHz 3.0GHz 128MB 155W
AMD EPYC™ 7313 16 32 Up to 3.7GHz 3.0GHz 128MB 155W
AMD EPYC™ 7303P 16 32 Up to 3.4GHz 2.4GHz 64MB 130W
AMD EPYC™ 7303 16 32 Up to 3.4GHz 2.4GHz 64MB 130W
AMD EPYC™ 72F3 8 16 Up to 4.1GHz 3.7GHz 256MB 180W
AMD EPYC™ 7203P 8 16 Up to 3.4GHz 2.8GHz 64MB 120W
AMD EPYC™ 7203 8 16 Up to 3.4GHz 2.8GHz 64MB 120W
AMD EPYC

Find industry-leading data center solutions powered by AMD EPYC™

Sign up to receive the latest data center news and server content

SUBSCRIBE NOW    CONTACT AMD

  1. MLN-201: SPECrate®2017_int_base comparison based on published scores from www.spec.org as of 10/27/2023. Comparison of published 1P AMD EPYC 7203P (70.8 SPECrate®2017_int_base, 120 Total TDP W, 8 Total Cores, $2635 Est system $, 207 est system W, https://www.spec.org/cpu2017/results/res2023q3/cpu2017-20230828-38848.html) is 1.62x the performance of published 1P Intel Xeon Bronze 3408U (43.7 SPECrate®2017_int_base, 125 Total TDP W, 8 Total Cores, $3074 Est system $, 251 est system W, https://www.spec.org/cpu2017/results/res2023q4/cpu2017-20230925-39034.html)  [at 1.96x the performance/system W] [at 1.89x the performance/system $]. AMD 1Ku pricing and Intel ARK.intel.com specifications and pricing as of 10/27/2023.  SPEC®, SPEC CPU®, and SPECrate® are registered trademarks of the Standard Performance Evaluation Corporation. See www.spec.org for more information. The system pricing and watt estimates are based on Bare Metal GHG TCO v9.60. Actual costs and system watts will vary.
  2. MLN-098B: SPECrate®2017_int_base comparison based on best performing systems published at www.spec.org as of 10/11/2023. Configurations: 2x AMD EPYC 7543 (567 SPECrate®2017_int_base, https://www.spec.org/cpu2017/results/res2021q4/cpu2017-20211011-29672.html , $7522 1Ku price total, 450W total TDP) versus 2x Intel Xeon Platinum 8358 (507 SPECrate®2017_int_base, https://www.spec.org/cpu2017/results/res2023q1/cpu2017-20230130-33812.html , $9214 1Ku price total, 500W total TDP) for 1.12x the performance at 1.36x the score per total CPU $ and 1.24x the performance/Watt. AMD 1Ku pricing and Intel ARK.intel.com specifications and pricing as of 10/11/23. SPEC®, SPEC CPU®, and SPECrate® are registered trademarks of the Standard Performance Evaluation Corporation. see www.spec.org for more information. 
  3. MLN-094B: SPECpower_ssj 2008 overall ssj_ops/watt comparison based on highest system results published as of 10/23/2023. Configurations: 2x AMD EPYC 7763 (64C) (25302 overall ssj_ops/watt, https://www.spec.org/power_ssj2008/results/res2022q3/power_ssj2008-20220617-01179.html) versus 2x Intel Xeon Platinum 8380 (40C) (13670 overall ssj_ops/watt, https://www.spec.org/power_ssj2008/results/res2022q4/power_ssj2008-20220926-01184.html) for 1.85x the performance per watt. SPEC® and SPECpower_ssj® are registered trademarks of the Standard Performance Evaluation Corporation. see www.spec.org for more information.
  4. GD-183: AMD Infinity Guard features vary by EPYC™ Processor generations. Infinity Guard security features must be enabled by server OEMs and/or Cloud Service Providers to operate. Check with your OEM or provider to confirm support of these features. Learn more about Infinity Guard at https://www.amd.com/en/technologies/infinity-guard
  5. EPYC-028: As of 2/2/22, of SPECpower_ssj® 2008 results published on SPEC’s website, the 55 publications with the highest overall efficiency results were all powered by AMD EPYC processors.  More information about SPEC® is available at http://www.spec.org.  SPEC and SPECpower are registered trademarks of the Standard Performance Evaluation Corporation. 
    Links to these 55 results are:
    1. http://www.spec.org/power_ssj2008/results/res2020q4/power_ssj2008-20200918-01047.html
    2. http://www.spec.org/power_ssj2008/results/res2020q4/power_ssj2008-20200918-01046.html
    3. http://www.spec.org/power_ssj2008/results/res2021q2/power_ssj2008-20210324-01091.html
    4. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200519-01031.html
    5. http://www.spec.org/power_ssj2008/results/res2021q1/power_ssj2008-20210309-01077.html
    6. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200407-01022.html
    7. http://www.spec.org/power_ssj2008/results/res2021q2/power_ssj2008-20210408-01094.html
    8. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200519-01034.html
    9. http://www.spec.org/power_ssj2008/results/res2021q2/power_ssj2008-20210413-01095.html
    10. http://www.spec.org/power_ssj2008/results/res2021q1/power_ssj2008-20210309-01078.html
    11. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200519-01032.html
    12. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200407-01023.html
    13. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200407-01025.html
    14. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200519-01033.html
    15. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200407-01024.html
    16. http://www.spec.org/power_ssj2008/results/res2021q4/power_ssj2008-20211001-01130.html
    17. http://www.spec.org/power_ssj2008/results/res2021q2/power_ssj2008-20210602-01106.html
    18. http://www.spec.org/power_ssj2008/results/res2021q2/power_ssj2008-20210602-01105.html
    19. http://www.spec.org/power_ssj2008/results/res2020q3/power_ssj2008-20200714-01039.html
    20. http://www.spec.org/power_ssj2008/results/res2020q1/power_ssj2008-20191125-01012.html
    21. http://www.spec.org/power_ssj2008/results/res2021q2/power_ssj2008-20210615-01111.html
    22. http://www.spec.org/power_ssj2008/results/res2020q3/power_ssj2008-20200714-01040.html
    23. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200324-01021.html
    24. http://www.spec.org/power_ssj2008/results/res2020q1/power_ssj2008-20191125-01011.html
    25. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200313-01020.html
    26. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200313-01019.html
    27. http://www.spec.org/power_ssj2008/results/res2020q1/power_ssj2008-20200310-01018.html
    28. http://www.spec.org/power_ssj2008/results/res2019q3/power_ssj2008-20190717-00987.html
    29. http://www.spec.org/power_ssj2008/results/res2019q3/power_ssj2008-20190717-00988.html
    30. http://www.spec.org/power_ssj2008/results/res2019q3/power_ssj2008-20190909-01004.html
    31. http://www.spec.org/power_ssj2008/results/res2019q3/power_ssj2008-20190717-00986.html
    32. http://www.spec.org/power_ssj2008/results/res2021q1/power_ssj2008-20210221-01066.html
    33. http://www.spec.org/power_ssj2008/results/res2019q3/power_ssj2008-20190717-00990.html
    34. http://www.spec.org/power_ssj2008/results/res2019q3/power_ssj2008-20190717-00985.html
    35. http://www.spec.org/power_ssj2008/results/res2020q3/power_ssj2008-20200728-01041.html
    36. http://www.spec.org/power_ssj2008/results/res2021q1/power_ssj2008-20210221-01063.html
    37. http://www.spec.org/power_ssj2008/results/res2019q3/power_ssj2008-20190716-00980.html
    38. http://www.spec.org/power_ssj2008/results/res2021q1/power_ssj2008-20210221-01064.html
    39. http://www.spec.org/power_ssj2008/results/res2021q1/power_ssj2008-20210221-01065.html
    40. http://www.spec.org/power_ssj2008/results/res2019q3/power_ssj2008-20190716-00982.html
    41. http://www.spec.org/power_ssj2008/results/res2021q1/power_ssj2008-20210223-01073.html
    42. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200407-01029.html
    43. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200407-01028.html
    44. http://www.spec.org/power_ssj2008/results/res2019q3/power_ssj2008-20190716-00981.html
    45. http://www.spec.org/power_ssj2008/results/res2019q4/power_ssj2008-20191203-01015.html
    46. http://www.spec.org/power_ssj2008/results/res2021q1/power_ssj2008-20210222-01068.html
    47. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200407-01026.html
    48. http://www.spec.org/power_ssj2008/results/res2021q1/power_ssj2008-20210223-01074.html
    49. http://www.spec.org/power_ssj2008/results/res2019q3/power_ssj2008-20190911-01005.html
    50. http://www.spec.org/power_ssj2008/results/res2021q1/power_ssj2008-20210222-01069.html
    51. http://www.spec.org/power_ssj2008/results/res2019q3/power_ssj2008-20190730-00994.html
    52. http://www.spec.org/power_ssj2008/results/res2021q1/power_ssj2008-20210222-01071.html
    53. http://www.spec.org/power_ssj2008/results/res2020q2/power_ssj2008-20200407-01027.html
    54. http://www.spec.org/power_ssj2008/results/res2019q3/power_ssj2008-20190717-00984.html
    55. http://www.spec.org/power_ssj2008/results/res2021q1/power_ssj2008-20210222-01072.html

  6. MLNTCO-010A:  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 EPYC™ SERVER VIRTUALIZATION and GREENHOUSE GAS EMISSIONS TCO ESTIMATION TOOL compares the 1P AMD EPYC™ EPYC_7453 (28c) and the 2P Intel® Xeon® Gold_6334 (8c) server solutions required to deliver 320 total virtual machines (VM), requiring 1 core and 8GB of memory per VM.  The analysis includes hardware components and may include virtualization software (SW) if selected.   


    SCENARIO PARAMETERS: This analysis compares an AMD EPYC 1P EPYC_7453 powered server with an Intel® Xeon® 2P Gold_6334 based server.  Due the large variation in these costs these components: Admin costs are not included as part of this analysis.  Real estate costs are not included as part of this analysis.   
     

    CORE ASSUMPTIONS:   
    Cost of Power @ 0.128 per kWh; Power per rack for server use 8 kW (kilowatts); PUE (power usage effectiveness) of 1.7; server rack size of 42RU. Each server has 1 hard drive. Server racks are considered pre-existing (no cost) for this analysis.  The server power in this analysis is modeled at 100% of TDP for all servers. 
     

    HARDWARE CONFIGURATIONS & COSTS: 
    The EPYC powered solutions requires 12 servers with the following configuration: 1 CPUs @ $1570 each; 4 x 64GB DDR4 DIMMs @ $136 each, in a 1 RU chassis @ $1750 ea.  The 12 EPYC_7453 servers have a total hardware acquisition cost of $50911.   
    The Intel based solutions requires 20 servers with the following configuration:  2 CPUs @ $2607 each; 8 x 16GB DDR4 DIMMs @ $49 each, in a 2 RU chassis @ $2025 ea. The 20 Gold_6334 servers have a total hardware acquisition cost of $160218. 
    The AMD EPYC solution has a 68% lower hardware acquisition cost.  
     

    POWER ONLY OPERATING COSTS: 
    AMD EPYC: Total solution power for the 12 AMD EPYC powered servers over the 3-years of this analysis is 174236kWh at a total cost of $22302.  
    Intel® Xeon®: Total solution power for the 20 Intel based servers over the 3-years of this analysis is 425030kWh at a total cost of $54404. 
    The AMD solution uses 59% less power at a 59% lower cost, which yields a 59% lower power-only OpEx. 
     

    HARDWARE ONLY TCO:  
    AMD total cost for 12 EPYC powered servers is $50911; total Power-Only OpEx is $22302 for a total estimated TCO of $73213  
    Intel® Xeon® total cost for 20 based servers is $160218; total Power-Only OpEx is $54404 for a total estimated TCO of $214622.   
    AMD has a $141409 lower or 66% lower TCO than the Intel solution.  
     

    VIRTUALIZATION TCO ASSUMPTIONS: 
    Virtualization software: VMware® vSphere Enterprise Plus w/ Production support with a user defined price of $6558.32 per Socket + Core with 3 year support. VMware® vSphere Enterprise Plus w/ Production support - 24x7 3yr support.   One software license is used or a processor (CPU) with up to 32 CPU cores.  If the CPU has more than 32 cores, one additional software license is required for every 32 core increment in that socket.  More information on VMware software can be found @ https://store-us.vmware.com/products/data-center-virtualization-cloud-infrastructure.html.   
     

    3-YEAR VIRTUALIZATION TCO: 
    The AMD solution requires 12 licenses at total cost of $78700 yielding a total 3-yr AMD Virtualization Solution TCO of $151913.  
    The Intel® Xeon® solution requires 40 licenses at a total cost of $262333 yielding a total 3-yr Virtualization Solution TCO of $476955.  
    The AMD solution requires 28 fewer or 70% less virtualization licenses and the AMD Virtualization TCO is $325042 less for a 68% lower TCO. 
     

    FIRST YEAR VIRTUALIZATION TCO: 
    AMD 1st year Virtualization Solution TCO is $137045; the Intel® Xeon® 1st year Virtualization TCO is $440685.  
    AMD 1st year TCO is $303640 less or 69% lower than Intel® Xeon®.   
    The AMD 1st year TCO per VM is $428; Intel® Xeon® is $1377.  
    The AMD 1st year TCO per VM is $949 less or 69% lower.  
     

    ENVIRONMENTAL: 
    The 1P AMD EPYC_7453 powered server solution uses ~174236kWh of electricity.  The Intel® Xeon® solution uses ~425030kWh.  The AMD solution saves ~250793 or 59% kWh of electricity over the 3-years of this analysis.  Leveraging this data, using the Country / Region specific electricity factors from the Carbonfootprint.com emissions factors sources, Feb 2023 update (United States was the country selected for this analysis with an average kgCO2e per kWh production fuel mix factor of 0.373138) and the United States Environmental Protection Agency.   'Greenhouse Gas Equivalencies Calculator', the AMD EPYC powered server saves ~93.58 Metric Tons of CO2 equivalents for the 3-years of this analysis, or 103.15 US tons which is 34.38 US tons per year This results in the following estimated savings based on United States data,  
    Greenhouse Gas Emissions Avoided of one of the following: 
         20 USA Passenger Cars Not Driven for 1 year; or 
         7 USA Passenger Cars Not Driven Annually; or 
         232267 Miles Driven by Avg Passenger Car; or 
    CO2 Emissions Avoided from: 
         10575 Gallons of Gasoline Not Used; or 
         103500 Pounds of Coal Not Burned in USA; or 
         18 USA Homes' Electricity Use for 1 year; or 
         6 USA Homes' Electricity Use Annually; or 
    Carbon Sequestered equivalent to: 
         1544 Tree Seedlings Grown for 10 years in USA; or 
         112 Acres of USA Forests in 1 year; or 
         37.43 Acres of USA Forests Annually 

    The CARBON FOOTPRINT COUNTRY SPECIFIC ELECTRICITY GRID GREENHOUSE GAS EMISSION FACTORS data used in this analysis was sourced at Carbonfootprint.com emissions factors sources, Feb 2023 update and can be found at:  https://www.carbonfootprint.com/docs/2023_02_emissions_factors_sources_for_2022_electricity_v10.pdf and the US EPA Greenhouse Gas Equivalencies Calculator used in this analysis was sourced on 03/03/2023 and can be found at https://www.epa.gov/energy/greenhouse-gas-equivalencies-calculatorVirtualization software pricing sourced online as of 03/06/2023.  Third party names are for informational purposes only and may be trademarks of their respective owners.  AMD CPU pricing based on 1KU price as of Sept 27, 2023. Intel® Xeon® Scalable processor data including pricing from https://ark.intel.com as of Sept 27, 2023.  Memory pricing sourced online from ATIC™ (use advanced search): http://www.atic.ca/ on 10/02/2023. Conversion to USD was at 1 CAD = 0.731663 USD from https://www.forbes.com/advisor/money-transfer/currency-converter/cad-usd/ on 10/02/2023. All pricing is in USD. 
    Results generated by: AMD EPYC™ Server Virtualization and Greenhouse Gas Emissions TCO Estimation Tool - v14.10