Beyond Balanced: Benchmarking Windows 11 Power Plans

Table of Contents

• What Power Plans are there in Windows?
• Setup and testing methodology
• Hypothesis and results
• Conclusion

In Windows, power plans provide a collection of settings that control how the system balances performance and power consumption. Depending on the workload and system configuration, selecting a different power plan could potentially improve performance or reduce power consumption.

While adjusting the power plan is relatively straightforward, the question is whether it actually results in a measurable performance difference. Is there a benefit to selecting a more performance-oriented power plan, or are the default settings already sufficient for most workloads?

This research tests the performance impact of the various power plans available in Windows 11 and determines whether adjusting the power plan provides a measurable benefit.

What Power Plans are there in Windows?

As mentioned in the introduction, a power plan is a collection of settings that controls how the system balances performance and power consumption. The power plan as we know it today was introduced with Windows Vista. However, even Windows XP included power schemes that could be used to manage settings such as display and hard drive timeouts.

One of the main drivers behind the introduction of these power plans was the significant increase in laptop adoption during the mid-2000s, which created a greater need to extend battery life. Today, power efficiency remains highly relevant, as laptops represent a significant portion of Windows-based endpoint deployments.

With a default installation of Windows, several power plans are available, each designed to balance performance and power consumption differently.

You can set the profiles via the user interface by going to Start Menu -> Control Panel -> System and Security -> Power Options.

This can also be done via the command line using the following commands.

powercfg /GETACTIVESCHEME

powercfg /SETACTIVE <GUID>

Please note that the Ultimate Performance power plan might not be available by default on all systems. If the power plan is missing, the following instructions explain how to enable it.

The Ultimate Performance power plan is the latest power plan available in Windows and was introduced with Windows 10 in 2018. Its main purpose is not necessarily to increase the maximum performance of the system. Instead, it is designed to reduce power-saving behaviour and the potential latency introduced when hardware transitions between different power states.

This can affect various components, including processor power management, disk idle behaviour, PCIe link-state power management, and peripheral power-saving settings. As a result, the potential benefit of Ultimate Performance might be more noticeable when a system frequently transitions between idle and active states rather than during a continuously active workload.

Source: Windows Forum

Why would you change this? This depends on the form factor you are using and the workload. In the case of a powerful workstation or a virtual machine, adjusting the power plan to High performance or Ultimate Performance could ensure that the system is always ready to deliver the maximum available performance at any given moment, without hesitation or (micro) stutters. In the case of a laptop or similar portable and mobile device, where mobility and battery life are more important, the Power saver profile might be more appropriate.

Looking at the available Microsoft guidance, no specific recommendations can be found that define which power plan should be used for a particular use case. Besides that, at the time of writing, there are no performance reports available from Microsoft that demonstrate the performance impact of the different power plans.

Although configuring the appropriate power plan is considered a best practice in many VDI optimization guidelines and most guides suggest selecting the highest performance plan available, there is limited research available that demonstrates the direct performance benefit. Therefore, it remains unclear how much of a measurable performance difference the various power plans actually make, particularly in EUC environments where performance is often desired over battery consumption.

Setup and testing methodology

For this research, an environment in Microsoft Azure is used to validate the various power plans. As multiple power plans are tested, each is included as a separate scenario in the research:

• Balanced
• High performance
• Ultimate Performance
• Power saver

Ten individual machines are deployed in Azure using the Standard_D4ds_v6 SKU, which provides 4 vCPUs and 16 GB of memory, combined with a 128 GB Premium SSD. Each machine runs its own benchmark, producing ten individual runs that are averaged for each scenario. Windows 11 25H2 is used as the operating system without any specific optimizations, with the exception of disabling Microsoft Defender. This ensures the focus is on the behaviour of the different power plans to be validated out of the box.

Please note that disabling Microsoft Defender is not recommended for any environment. However, this is a common practice in controlled benchmarking environments to prevent background Defender processes from introducing unexpected performance variations and to ensure consistent and repeatable results.

The benchmark solution used for this research is OBUX release v0.1.5.94, which is initiated through an on-premises RDP connection to the Azure-based machines. Each machine runs the individual test, resulting in 10 OBUX results per scenario.

After changing the power plan between scenarios, each machine is rebuild before starting the benchmark to ensure consistent test conditions. This entire process is fully automated, following the standard testing methodology used at GO-EUC. This research deliberately excludes standby and hibernation and will not focus on P and C state control of the power plans.

Hypothesis and results

As applying at least the High performance power plan is considered a best practice when optimizing a golden image, it is expected that this will provide a measurable performance benefit, resulting in better scores compared to the other power plans.

As OBUX is the primary benchmark used in this research, let’s start with the results it produces. The OBUX System Score is the best indicator for identifying a direct performance difference, as it includes various calculations designed to stress the system. One of the main benefits of using OBUX is that it allows the results from the different scenarios to be easily compared, providing a clear indication of the relative performance differences between the power plans.

There is a difference between the various power plans, especially when compared to Power saver. However, the scores are relatively close and do not show a significant increase in performance. It is also noticeable that the overall System Score for the Ultimate Performance power plan is slightly lower compared to High Performance, which is not entirely expected. The individual runs show a higher variance in scores, which affects the average score.

The UX Score provides an indication of whether the different power plans have a measurable effect on the user experience and responsiveness of the system.

Although the difference is more noticeable in the System Score, there is minimal to no difference in the UX Score. This indicates that switching between the different power plans has no measurable impact on the responsiveness of the system during the OBUX benchmark.

During the benchmarks, various system performance metrics are also collected, allowing us to dive deeper into the results and determine whether the different power plans have an impact that is not directly reflected in the OBUX scores.

CPU utilization is nearly identical across all scenarios, indicating that changing the power plan does not have a measurable impact on CPU utilization during the OBUX benchmark.

CPU Queue Length is a metric that is not often used in GO-EUC research. For a better understanding, this metric provides insight into the amount of work waiting to be processed by the CPU. In this scenario, it is an interesting metric, as a power plan could potentially reduce CPU performance and consequently result in a higher CPU queue.

However, during the OBUX benchmark, there is no significant difference in CPU Queue Length between the various power plans.

The memory footprint is identical across all scenarios, indicating that the different power plans have no measurable impact on memory consumption.

Disk idle time is another metric that is not commonly used in GO-EUC research. However, in this case, it provides insight into whether a specific power plan is more aggressively idling the disk. The higher the percentage, the more time the disk spends in an idle state.

Looking at the results, the differences between the various power plans are negligible. This indicates that, within the scope of this research, changing the power plan does not have a measurable impact on disk idle time.

Conclusion

To answer the main question: Is there a benefit to selecting a more performance-oriented power plan? Not really.

Based on the collected data, there is a slight increase in performance when changing the power plan from Balanced to High performance or Ultimate Performance, but don’t expect a significant boost in performance.

However, it is important to fully understand the context of this research before drawing any definitive conclusions. The scope and methodology of this research focus on benchmarking the system using OBUX. OBUX is designed to validate the maximum performance and user experience of a system by performing computational calculations to stress the system while also measuring responsiveness through interactions with baseline applications.

The execution of this research is designed to run and collect data as efficiently as possible. This means that the system is never in a fully idle state before or during the test. It is also important to note, as stated in the setup and scenario, that P and C states are not in scope of this research. As power plans can influence the behaviour of certain system components during idle or low-utilisation periods, these conditions were not covered within the scope of this research.

If you are an administrator of a VDI environment, a knowledge or power worker, a graphic designer, or even a gamer, there is no harm in adjusting the power plan. However, based on the results of this research, you should not expect a direct boost in performance simply by changing from Balanced to High performance or Ultimate Performance.

It is important to note that when searching for information about Windows power plans, there are various discussions over the internet, particularly within the gaming community, claiming that adjusting or tweaking the power plan can directly improve gaming performance and result in higher frame rates.

However, the gaming workload and testing methodology are different from the scenarios covered in this research. Based on the collected data, it is not possible to confirm these claims. This would require additional testing with gaming-specific workloads and should be considered for future GO-EUC research. With this context, the conclusion of this research still stands. Within the tested scenario, changing the power plan does not provide a measurable performance benefit. However, the behaviour of the different power plans during idle and low-utilisation conditions has not been taken into account and might require further investigation from a GO-EUC perspective.

Do you adjust your power plan in your environment, and do you think this research requires a follow-up?

Photo by Zac Wolff on Unsplash