Intel To Utilize AI To Make Next-Gen Meteor Lake CPUs More Power Efficient & Faster

30.08.2023 - 12:27 / wccftech.com / Meteor Lake / Ai

Intel Meteor Lake CPUs aren't just going to make use of AI to make user experiences better but also improve its own internal capabilities.

There have been various ways to control active power and idle power states on past generations of Intel CPUs, reports PCWorld. With Meteor Lake CPUs, Intel is infusing some key AI elements not just to make computing experiences better and more intuitive, but also to help the chip with its internal power management.

So the process remains the same, how fast a CPU gets the work done and how fast it can return to its idle state. With AI, this process can be accelerated, leading to lower power waste while the CPUs are working. This is the first-of-its-kind technology that will be integrated first on Meteor Lake CPUs and all future chips.

With Meteor Lake CPUs, AI will help determine what is the right amount of power to be fed to the CPU for a specific workload. The power will translate into frequency and that would translate into a responsive PC which can yield better performance when required. This is a crucial aspect for CPUs aiming at the mobility & laptop segment where battery times play a huge role. The AI will become an integral part of the DVFS (Dynamic Voltage & Frequency Scaling).

According to Intel, AI can help improve the system responsiveness by 35% and in some cases, even higher. Each iteration of future CPUs will benefit from this AI fusion which is part of Intel's next-gen disaggregated lineup. And AI can also determine when is the right moment to put the CPU back in an idle state, leading to 15% power savings.

In the following example, Intel shows the SOC Power of an upcoming Meteor Lake CPU running with and without AI. You can see that the Non-AI chip consumes more power & also takes longer when returning to an idle state.

The company has already made strides by offerings its first hybrid architecture solutions in the form of 12th Gen Alder Lake and 13th Gen Raptor Lake CPUs. These chips include P-Core that's designed to handle workloads that require more performance and E-Cores that are designed for efficient task and multi-threaded processing. Meteor Lake will continue this path with its P-Core (Redwood Cove) and E-Core (Crestmont) architectures.

As for the AI itself, Intel's Efraim Rotem of Design Engineering Group explains that in the current iteration, the AI has been trained on specific scenarios across a diverse range of workloads but at the current moment, the AI isn't learning "dynamically" from user-based interactions on the fly. It might be something that we see in future generations but for now, it will be designed to handle a specific set of use cases and still deliver the benefits to end users. Intel is also working on different AI models for