How much power does ARM CPU use?

ARM CPUs are designed to be power efficient while still delivering high performance. The amount of power an ARM CPU uses depends on the specific chip architecture, clock speed, workload, and power saving features enabled.

Key Factors That Determine ARM CPU Power Consumption

There are several key factors that determine how much power an ARM CPU will draw:

• CPU architecture – More advanced ARM architectures like ARMv8 are more power efficient than older designs like ARMv7.
• Manufacturing process – Smaller manufacturing processes like 7nm allow for lower voltage operation.
• Clock speed – Higher CPU clock speeds require more power.
• Core count – More CPU cores consume more total power but can distribute work efficiently.
• Workload – Light workloads consume less power than heavy multithreaded workloads.
• Power saving features – Dynamic voltage and frequency scaling and idle power states can reduce power draw.

ARM CPU Power Consumption Ranges

Here are some approximate power consumption figures for different types of ARM CPUs:

• Simple microcontroller-class ARM cores may use as little as 10-20mW.
• Efficient mobile ARM cores like Cortex-A55 can use around 250mW at peak performance.
• High performance mobile CPUs like Snapdragon 865+ use 2-5W in normal use cases.
• Server-class ARM CPUs like Ampere Altra draw 30-40W under load.
• The highest performance ARM chips for HPC can consume over 200W when fully utilized.

As you can see, there is a very wide range – from milliwatts for simple embedded cores to over 100W for server chips. But in general, ARM designs are engineered for power efficiency compared to x86 CPUs.

Factors That Reduce ARM CPU Power Consumption

There are a number of architectural optimizations used in ARM CPU design to reduce power consumption:

• Reduced Instruction Set (RISC) – RISC architectures require fewer transistors which reduces power draw.
• Advanced Manufacturing Nodes – Smaller transistors can operate at lower voltages to reduce power.
• Big.LITTLE – Mixing high performance and efficient CPU cores allows optimal power scaling.
• Heterogeneous Compute – Offloading work to specialized processors like GPUs and NPUs.
• Dynamic Voltage & Frequency Scaling – Clock speed and voltage can scale based on workload.
• Idle Power States – Parts of the CPU can be powered down when not needed.

Measuring ARM CPU Power Consumption

To accurately measure the power draw of an ARM CPU, you will need a few tools:

• Multimeter – For measuring current draw at the board level.
• Power monitor – Used between the power supply and board to measure overall system power.
• Logic analyzer – Provides power draw over time at the SoC level.
• Thermal camera – Visualizes hot spots and heat dissipation issues affecting power.
• Embedded power sensors – Some ARM SoCs feature on-die power monitoring capabilities.

Carefully instrumenting and measuring power at each level provides a detailed view of ARM CPU power consumption under different workloads and configurations.

Typical ARM CPU Power Consumption Profiles

Looking at a typical modern ARM-based smartphone SoC like the Snapdragon 865, we can see variations in CPU power consumption under different scenarios:

• Idle state power draw – ~10mW
• Light use (web browsing) – ~250mW average
• Gaming – 2-3W sustained power
• Heavy multithreaded workload – ~4-5W power

For a high performance ARM server CPU like Ampere Altra:

• Idle state – ~5W
• Moderate compute tasks – ~15-30W
• All cores loaded – Up to 200W power draw

Measuring the power at different levels of utilization demonstrates ARM’s ability to be extremely power efficient under light workloads while still scaling up to deliver high performance when needed.

Optimizing ARM CPU Power Efficiency

There are also steps developers can take to optimize power efficiency when designing using ARM CPUs:

• Take advantage of heterogeneous computing by offloading to GPUs, NPUs, and DSPs.
• Minimize memory accesses and unnecessary data movement.
• Ensure workloads can scale across all cores to maximize performance per watt.
• Use power efficient languages like C instead of managed languages.
• Profile power consumption and look for hot spots.
• Use low power modes and turn off peripherals when not needed.

Efficient coding practices combined with ARM’s architectural advantages enables industry-leading power efficiency for embedded and mobile applications.

Conclusion

ARM CPUs deliver an impressive combination of high performance and power efficiency. Simple ARM cores can sip power at milliwatt levels while advanced multi-core server chips can still draw 100+ watts under full load. But efficient RISC architecture, advanced manufacturing, and intelligent power management allows ARM processors to often beat alternative architectures like x86 in performance per watt – a key metric for embedded and mobile applications.