Azure compute unit (ACU)

Applies to: ✔️ Linux VMs ✔️ Windows VMs ✔️ Flexible scale sets ✔️ Uniform scale sets

Note

This page is no longer being maintained and is planned for archival on 12/16/2024.

Azure is no longer publishing ACUs since the metric has limited ability to inform users of the expected performance of a virtual machine across various attributes. For the most accurate results on a specific virtual machine, Azure recommends users run their workload(s) on that virtual machine to verify performance.

The concept of the Azure Compute Unit (ACU) provides a way of comparing compute (CPU) performance across Azure SKUs. The Azure Compute Unit enables you to easily identify which SKU is most likely to satisfy your performance needs. ACU is currently standardized on a Small (Standard_A1) virtual machine being 100 and all other SKUs then represent approximately how much faster that SKU can run a standard benchmark.

*ACUs use Intel® Turbo technology to increase CPU frequency and provide a performance increase. The amount of the performance increase can vary based on the virtual machine size, workload, and other workloads running on the same host.
**ACUs use AMD® Boost technology to increase CPU frequency and provide a performance increase. The amount of the performance increase can vary based on the virtual machine size, workload, and other workloads running on the same host.
***Hyper-threaded and capable of running nested virtualization.
****AMD Simultaneous multithreading technology.

Important

The ACU is only a guideline. The results for your workload may vary.

SKU Family ACU \ vCPU vCPU: Core
A1_v2 - A8_v2 100 1:1
A2m_v2 - A8m_v2 100 1:1
B Varies 1:1
D1 - D14 160 - 250 1:1
D1_v2 - D15_v2 210 - 250* 1:1
DS1 - DS14 160 - 250 1:1
DS1_v2 - DS15_v2 210 - 250* 1:1
D_v3 160 - 190* 2:1***
Ds_v3 160 - 190* 2:1***
Dav4 230 - 260** 2:1****
Dasv4 230 - 260** 2:1****
Dv4 195 - 210 2:1***
Dsv4 195 - 210 2:1***
Ddv4 195 -210* 2:1***
Ddsv4 195 - 210* 2:1***
E_v3 160 - 190* 2:1***
Es_v3 160 - 190* 2:1***
Eav4 230 - 260** 2:1****
Easv4 230 - 260** 2:1****
Ev4 195 - 210 2:1***
Esv4 195 - 210 2:1***
Edv4 195 - 210* 2:1***
Edsv4 195 - 210* 2:1***
F2s_v2 - F72s_v2 195 - 210* 2:1***
F1 - F16 210 - 250* 1:1
F1s - F16s 210 - 250* 1:1
M 160 - 180 2:1***
Mv2 240 - 280 2:1***
NVv4 230 - 260** 2:1****

Processor model information for each SKU is available in the SKU documentation. Optimal performance may require the latest virtual machine images (OS and virtual machine generation) to ensure the latest updates and fastest drivers.

Virtual machine series retiring

The following virtual machine series are retiring on or before August 31, 2024.

SKU Family ACU \ vCPU vCPU: Core Retirement Date
A0 50 1:1 August 31, 2024
A1 - A4 100 1:1 August 31, 2024
A5 - A7 100 1:1 August 31, 2024
A8 - A11 225* 1:1 August 31, 2024

Performance consistency

Azure virtual machine sizes typically run with maximum performance on the hardware platform they're first released on. Azure may place controls on older Azure virtual machines when run on newer hardware to help maintain consistent performance for our customers even when the virtual machines run on different hardware. For example:

  • D, E, and F series virtual machines may have the processor frequency set to a lower level when running on newer hardware to help achieve better performance consistency across hardware updates. (The specific frequency setting varies based on the processor the virtual machine series was first released on and the comparable performance of the current hardware).
  • A series virtual machines use an older model based on time slicing newer hardware to deliver performance consistency across hardware versions.
  • B series virtual machines are burstable and use a credit system (described in their documentation to achieve expected performance.

These different processor settings for virtual machines are a key part of Azure's effort to provide consistent performance and minimize the impact of changes in underlying hardware platform outside of our customer’s control.

More information

For more information about the different sizes, see: