Current laptops with Intel Core Ultra Series 2 processors are based on a design of hybrid fleas specifically focused on energy efficiency. The neuronal treatment unit (NPU), used for the first time in consumption systems, plays a central role here. This computer unit dedicated for the tasks of AI relieves the CPU and the GPU of processes based on inference such as image recognition, language processing or modeling.
While the CPU must have assumed many of these tasks in conventional systems, the NPU allows a significantly more differentiated load distribution. This reduces the load of the average system and significantly reduces energy needs. Like numerous NPU calculations can be carried out at a low clock frequency and in parallel, the energy balance is considerably improved compared to architectures purely based on the processor or the GPU.
Energy saving components in Intel Core Ultra
Intel Core Ultra V models in particular combine four performance nuclei with four efficiency nuclei and a dedicated NPU to form a computer unit on several levels. Cores P take care of critical performance tasks, while electronic cores and the NPU remain continuously active in the background and run routine processes and AI functions with low-power needs.
Mark Hachman / IDG
Intel Arc Graphics Intel ARC also play a role in this context: it allows a video decoding accelerated by the equipment and a display with high graphic intensity without an additional dedicated GPU, which relieves the cooling system and reduces overall energy consumption. The NPU offers up to 48 summits of the computing power with minimum energy consumption. This benefits AI applications and AI functions as well as users, because the energy needs of laptops can be considerably minimized.
Intel
Microsoft energy saving mechanisms under Windows 11
In addition to the hardware platform, new energy saving strategies have been implemented with Windows 11. “Management of user interaction processor” analyzes user activity in real time. If no interaction via the keyboard, the mouse or the touchpad is detected, the system automatically erases processor performance without interrupting the reading or the presentations of active media. In addition, the “adaptive energy saving” function also activates the energy saving mode regardless of the state of the battery, provided that the system of load and use of the system allows it.
Sam Singleton
In both cases, the NPU can ensure that the functions supported by AI remain active in the background without having a negative impact on the energy balance. The AI also balances the priorities in the background, for example by delaying the synchronization of the clouds or the adaptive process rest.
HP omnibook and other co -pilot models in comparison
Devices such as the HP OMNIBook X line already incorporate technology system. In combination with an Intel Core Ultra 7 258V GPU and an Intel ARC 140V GPU, the NPU allows locally executed features such as Windows studio effects or AI functions in HP Ai Compagne without much drainage of the battery. Many other models also reach battery times of more than 24 hours in mixed operation thanks to the use of NPUs. Models such as the surface laptop 6 or the PRO 10 surface incorporate an NPU dedicated directly into the Intel Ultra Soc nucleus, supplemented by high -performance processor nuclei and integrated graphics.
Other compatible devices are also based on the concept of co -pilot, which combines powerful NPUs with intelligent energy management. Devices such as the Galaxy Book with RTX 4050/4070 or the Surface Pro 10 with Intel Core Ultra 7 demonstrate these possibilities. In practice, this means that even when language translation, background blur or image optimization is actively used, energy consumption remains low.
Optimization based on AI software and unloading
A significant contribution to energy savings is made by moving the workloads with high calculation intensity at the NPU on the software side. Applications such as Zoom, Adobe Premiere Pro or Amuse are increasingly using interfaces based on ONNX execution time to unload AI processes such as image generation, object monitoring or audio filters in the NPU.
Adobe
This reduces the energy needs of the CPU, which is particularly visible for long periods of use in video conferences or creative applications. The NPU is accessible via standardized interfaces such as DirectML and Intel and AMD platforms, which have native integration in the ONNX execution time. The reduction in the load resulting from the main processors makes a decisive contribution to the more uniform load distribution and therefore to a longer battery life.
Interaction of CPU, GPU and NPU in practice
In modern notebooks, the CPU, the GPU and the NPU work as a dynamic treatment trio. While the CPU continues to control the operating system and general applications, the GPU supports graphic tasks or parallelized IT operations. The NPU focuses on dedicated AI processes and allows continuous treatment with low energy consumption. Windows 11 attributes these tasks specifically and continuously assesses which unit is the most effective for execution.
IDG / Mark Hachman
This means that recurring tasks such as transcription of speech, recognition of people or background noise filters can be treated directly on the NPU. This does not only reduce energy consumption, but also reduces the temperature of the system, which allows lighter cooling systems and therefore more compact and lighter laptop conceptions.
Local treatment instead of unloading the cloud
The local execution of the AI workloads on the NPU replaces the usual cloud access in many cases. This means that image analyzes, language models or layout suggestions no longer have to be calculated online, but fully run on the device. This not only reduces latency, but also avoids the activity of the unnecessary network. This is another factor that reduces energy consumption.
At the same time, the availability of these functions increases even without network connection, for example on the train or on a trip. The battery life then benefits in two ways: thanks to a lower calculation load on the CPU and the GPU and by the Wi-Fi or LTE / 5G activity.
Windows 11 shows the use of the NPU in the task manager for the first time
Microsoft has expanded the task manager for the control and transparency of this new architecture. In addition to the CPU, the GPU and the RAM, the use of NPU is now also displayed as a distinct measured value. This allows users to understand how much their AI applications really benefit from dedicated equipment.
For developers, the ONNX runtime in combination with the Windows performance analyzer also offers detailed diagnostic functions which can be used to specifically analyze inference times, operator load and load curves. This allows refined optimization for maximum energy gain and a minimum execution time.
Sam Singleton
The battery life as the new reference for PCs IA
Although attention has been focused for a long time on the computing power and the size of the model, there is now a paradigm shift. The actual execution time of a device is becoming more and more the most important quality criterion for optimized notebooks in AI. Modern AI notebooks reach video reading times of more than 26 hours in realistic conditions, a value that would be almost impossible to achieve without energy distribution supported by the NPU.
At the same time, the combination of an adaptive energy saving mode, local IA unloading and intelligent load controls opens up new possibilities for mobile applications where food is not always guaranteed.
Conclusion: Save energy with specialized AI equipment
The integration of UNPs into current notebooks of notebooks not only marks a technological advance in terms of AI performance, but also allows a lasting reduction in energy consumption thanks to the sharing of intelligent tasks for the first time. In combination with the new Windows 11 energy saving functions, the result is a platform that works not only more quickly in daily use, but also more effectively. For users, this means a longer battery life, less heat heat, quieter systems and better overall balance between performance and mobility, without sacrificing modern AI functions.
This article originally appeared on our PC-Welt sister publication and was translated and located in German.
