Efficiency Cores

The processor is the component that carries out the actual computing steps in a computer or phone. Today it no longer consists of a single computing unit, but of several, which are called cores. Each core can process a task independently. In many current chips, however, these cores are not all built the same. Some are designed for speed, others for low power consumption. The power-saving ones are called Efficiency Cores, often abbreviated as E-Cores.

Why chips today have two types of cores

A fast core consumes power disproportionately. Double the computing performance doesn’t cost double, but often four to five times the energy. Conversely, a core that works somewhat slower gets by on very little power. Manufacturers exploit exactly this relationship when they build both types into the same chip.

The practical benefit shows up in laptops and smartphones. Most of the time, after all, a device isn’t doing anything demanding: it’s checking mail, keeping the clock updated, downloading updates in the background. For such tasks, a high-performance core is grossly oversized. If all of this runs on Efficiency Cores, the battery lasts noticeably longer, and the fan stays quiet.

A second reason is space. An Efficiency Core takes up only a fraction of the chip area of a fast core. So the same area can fit four power-saving cores instead of one fast one. When many tasks run simultaneously, that’s the better trade-off: four slow cores together accomplish more than one fast one.

How work is distributed across the cores

The distribution is handled by the scheduler, the part of the operating system that decides which program gets to compute where at any given moment. It observes how much computing time a task needs and whether a human is waiting for the result. A video editing program in the foreground ends up on the fast cores. A backup service in the background moves to the Efficiency Cores.

For this to work reliably, the processor continuously sends hints to the operating system. Intel calls this technology Thread Director; Apple and Qualcomm use similar approaches. Without such feedback, the system is just guessing. Older versions of Windows therefore initially, and accidentally, pushed games onto the power-saving cores, which noticeably cost performance.

Technically, Efficiency Cores are often built more simply. They forgo elaborate circuitry that reorders instructions ahead of time, and they run at lower clock speeds. Both save power but make them slower per second. The instructions they understand, however, are the same. A program therefore doesn’t notice which type of core it’s currently running on.

Where E-Cores show up, in devices and specs

The principle was invented in the mobile phone sector. ARM introduced it in 2011 under the name big.LITTLE, and it has since found its way into practically every smartphone. Apple adopted it in 2020 for its Mac chips, Intel in 2021 with the Alder Lake processor generation. A chip like the Apple M4 has four fast and six power-saving cores.

When buying a computer, you’ll encounter this in the technical specifications. There it might say something like “8 P-Cores + 16 E-Cores”. The total number of cores alone says little. Two processors each with 24 cores can differ substantially, depending on the ratio between the two types.

A common misconception is that Efficiency Cores are merely a stopgap for weak devices. In servers and in data centers for artificial intelligence, often the opposite is true. There, what counts is how much work gets done per watt of power. Chips with very many power-saving cores are frequently the more economical choice for such tasks.