AMD Core Performance Boost for HP Zbook

Read, enable, or disable the AMD Core Performance Boost BIOS setting on supported HP systems with PowerShell.

---

Introduction

AMD processors can temporarily run above their base clock when there is enough power and thermal headroom. On supported HP systems, the BIOS setting that controls this behavior is called AMD Core Performance Boost. Enabling it can make the system noticeably faster, especially in short or lightly threaded workloads, but the extra performance also brings more heat, power use, and fan activity.

This project is a small PowerShell tool for reading and changing that setting on supported HP commercial PCs. It was written for a real deployment need and deliberately focuses on this one BIOS option.

Tested configuration

The script and benchmarks were tested on an HP ZBook 8 G1ak with an AMD Ryzen AI 5 PRO 340 processor and BIOS version 01.05.04.

For this processor, AMD lists:

• A 2.0 GHz base clock.
• A maximum boost clock of up to 4.8 GHz on the full Zen 5 cores.
• A maximum clock of up to 3.4 GHz on the compact Zen 5c cores.
• A maximum operating temperature (Tjmax) of 100 °C.

The 4.8 GHz figure is a peak single-core boost target, not a speed that every core should sustain. AMD explains that the clock reached in practice depends on the workload, system cooling, power limits, BIOS design, drivers, and operating system updates. Likewise, 100 °C is the processor's published thermal ceiling; it should not be treated as proof that this HP notebook will wait until exactly 100 °C before reducing performance.

Real-world observations

These are my field results from a production notebook, not results from a clean benchmark machine. Most measurements were taken while connected to AC power, followed by a separate battery-power comparison with boost enabled. The notebook is joined to an enterprise environment and runs the organization's normal security, endpoint-management, monitoring, and background applications. Those services consume resources and can affect scores, clock behavior, and idle temperatures.

For that reason, the numbers below should be used to compare boost enabled versus disabled on this system. They should not be compared directly with reviews, clean Windows installations, or the HP factory image.

Idle temperatures were recorded after startup and 15 minutes without a deliberate workload. Windows Task Manager was used for the visible clock reading, while CPU-Z supplied the benchmark and stress workloads.

AMD Core Performance Boost	CPU temperature after 15 minutes at idle	CPU speed observed in Task Manager
Disabled	Approximately 45–47 °C	Capped at approximately 2.0 GHz
Enabled	Approximately 61–65 °C	Boosted up to approximately 3.7 GHz

CPU-Z benchmark results on AC power

The AMD Ryzen AI 5 PRO 340 contains three full Zen 5 cores and three compact Zen 5c cores, with simultaneous multithreading providing six threads in each group. The results below use CPU-Z's P-Core label for the Zen 5 group and E-Core for the Zen 5c group.

Boost setting	CPU-Z test group	Threads	Single-thread score	Multi-thread score	Observed loaded speed	Maximum observed temperature
Disabled	All cores	12	235.6	2133.0	Approximately 2.0 GHz	50–51 °C
Disabled	P-Core / Zen 5	6	293.6	1207.6	Approximately 2.0 GHz	Not recorded
Disabled	E-Core / Zen 5c	6	295.0	1204.8	Approximately 2.0 GHz	Not recorded
Enabled	All cores	12	502.0	3724.1	Up to approximately 4.0 GHz	70 °C
Enabled	P-Core / Zen 5	6	718.5	2826.2	Up to approximately 4.4 GHz	70 °C
Enabled	E-Core / Zen 5c	6	514.6	2101.5	Not recorded	Not recorded

On this notebook, enabling boost improved the all-core multi-thread CPU-Z score by roughly 75% and the single-thread score by roughly 113%. The biggest gain appeared in the P-Core / Zen 5 group. During both the all-core and P-Core stress tests, the reported temperature levelled off at about 70 °C.

That 70 °C plateau does not by itself confirm thermal throttling. AMD specifies 100 °C Tjmax, but HP may apply a lower temperature target, power limit, or fan policy at the platform level. Confirming the exact limiting factor would require sensor data showing a thermal- or power-limit flag.

CPU-Z benchmark results on battery power

A second test was run after disconnecting AC power while leaving Core Performance Boost enabled. The same managed Windows environment and CPU-Z test groups were used.

CPU-Z test group	Threads	Single-thread score	Multi-thread score
All cores	12	500.0	3381.2
P-Core / Zen 5	6	461.3	2226.6
E-Core / Zen 5c	6	384.9	1589.8

During the continuous CPU-Z stress test, the 12-thread score settled at approximately 3520 and the CPU clock levelled off at about 3.0 GHz. A separate six-thread stress run produced a score of approximately 2195. CPU-Z stress scores are live workload readings, so they should not be treated as directly equivalent to the completed benchmark scores above.

Compared with the boost-enabled AC results, the all-core multi-thread benchmark score was about 9% lower on battery. The targeted P-Core results fell by roughly 36% in single-thread and 21% in multi-thread performance, while the E-Core results were about 25% lower. The all-core single-thread result was nearly unchanged, which may reflect preserved short-burst performance or normal scheduling variation. Together with the sustained 3.0 GHz clock, these results show that boost remains active on battery and are consistent with a more restrictive sustained power budget than on AC power.

My practical conclusion

Core Performance Boost delivers a clear advantage in raw CPU performance. For users running sustained, CPU-heavy workloads, the additional speed can be worthwhile. The tradeoff is equally visible: the processor runs warmer, the cooling system works harder, and fan noise is more noticeable.

For everyday office work, the decision is less straightforward. Applications such as email, web browsers, document editing, and video meetings may benefit from short bursts of extra performance, but they are unlikely to use the full gains seen in CPU-Z. The battery results also show that the notebook already reduces sustained boost when AC power is removed. In that setting, a cooler and quieter notebook may have a greater effect on the end-user experience than the additional benchmark performance.

The notebook's cooling design also matters. Its air intake is on the bottom, so paper, soft materials, or an uneven surface can partially block the airflow. A thermally induced shutdown has already been encountered under real-world conditions with restricted intake airflow. Restricted airflow can occur during normal enterprise use and should be considered when evaluating the additional heat produced with boost enabled.

Based on this test, I would enable boost for users who genuinely need sustained CPU performance and can keep the notebook on a firm, unobstructed surface. For typical office users, I would lean toward leaving it disabled when lower temperatures, quieter operation, and predictable behavior matter more than maximum CPU throughput. This is a personal conclusion from one managed notebook and workload mix, not a universal recommendation.

My results are also recorded as the first entry in BENCHMARKS.md. If you test another system, please add a separate entry there and submit it through a pull request. That file includes the CPU-Z link and a simple contribution template.

Applying the BIOS change

One practical detail showed up during repeated testing: after enabling boost, a normal Windows restart was not enough to make the new behavior active, even though the BIOS value read back correctly. Boost only became active after Windows was shut down completely and the notebook was powered on again.

Disabling boost did appear to work after a restart. Even so, the script asks for a complete shutdown in both directions. Using one consistent process is safer and avoids reporting success before the firmware behavior has actually changed.

What the script does

• Reads the current BIOS value and works with the value names exposed by the local firmware.
• Toggles the setting interactively when run without an action switch.
• Supports explicit -Enable and -Disable actions for repeatable deployment.
• Supports -Silent for unattended use and -DryRun for previewing a change.
• Reads the value again after writing it, so a failed BIOS update is not silently reported as successful.
• Returns a structured result object that can be captured by deployment tooling.

Before you run it

• A supported HP desktop, notebook, or workstation with an AMD processor that supports AMD Performance Boost.
• Windows PowerShell running as Administrator.
• HP Client Management Script Library (HPCMSL).
• Firmware that exposes the AMD Core Performance Boost setting.

If HPCMSL is missing, the script prepares the Windows PowerShell package-management components and installs the module from the PowerShell Gallery. In a managed environment, you can also deploy HPCMSL through your normal endpoint-management process before running the script.

Important

A BIOS setup password can block the change. Password handling is not currently implemented, so the script cannot update a protected BIOS setting.

Running the script

Open an elevated Windows PowerShell session in this project folder.

PowerShell execution policy

Choosing and enforcing the PowerShell execution policy is the responsibility of the organization's administrator. This project does not change the configured policy or require it to be bypassed.

Get-ExecutionPolicy -List

Review the effective policy before running the script. Group Policy can override local settings, so use your organization's approved process if script execution is blocked.

Where organizational policy explicitly permits it, Windows PowerShell also supports the process-level -ExecutionPolicy Bypass option:

powershell.exe -NoProfile -ExecutionPolicy Bypass -File .\AmdCorePerformanceBoost.ps1

This option applies only to that PowerShell process. It is shown as an available administrative mechanism, not as a recommendation or a requirement of this project.

Interactive mode

.\AmdCorePerformanceBoost.ps1

The script displays the current value, proposes the opposite value, and asks before making the change.

Choose a state explicitly

.\AmdCorePerformanceBoost.ps1 -Enable
.\AmdCorePerformanceBoost.ps1 -Disable

These commands do not show confirmation or shutdown prompts, which makes them suitable for repeatable automation. If the requested value is already set, the script leaves it unchanged.

Unattended mode

.\AmdCorePerformanceBoost.ps1 -Silent
.\AmdCorePerformanceBoost.ps1 -Enable -Silent
.\AmdCorePerformanceBoost.ps1 -Disable -Silent

Without -Enable or -Disable, -Silent toggles the current value without confirmation or shutdown prompts. You can still combine it with an explicit action when that makes a deployment command easier to read, although explicit actions are already non-interactive. Every mode returns a structured result. After a successful change, shut Windows down completely and power the computer on again.

Preview a change

.\AmdCorePerformanceBoost.ps1 -Disable -Silent -DryRun

-DryRun reports what the script would do without writing to the BIOS.

Automatic HP module installation

There is no separate installation switch. If HPCMSL is unavailable, the script installs it before trying to read the BIOS. This requires 64-bit Windows PowerShell 5.1 and access to the PowerShell Gallery.

Result statuses

Status	Meaning
Changed	The setting was changed and read back successfully. Shut down Windows completely, then power the computer on again to apply it.
Unchanged	The requested state was already set, or the current state could not be safely classified.
Cancelled	Interactive confirmation was declined.
DryRun	The requested action was previewed; nothing was changed.

Safety notes

• Test on one device before broad deployment.
• Keep AC power connected while changing firmware settings.
• Do not interrupt the system while a BIOS write is in progress.
• After a successful change, perform a complete shutdown rather than a restart.
• Use -DryRun first when deploying the script unattended.

Scope and limitations

This tool changes one BIOS setting. It does not update the BIOS, tune the GPU, overclock the processor, or promise a particular temperature or performance result. HP can expose different values and behavior across models and BIOS releases, so test on a small number of systems before wider deployment.

License

This project is licensed under the MIT License. You may use, copy, modify, distribute, subject to the license terms.