Arm has unveiled its latest processor designs for smartphones. These include the Cortex-A8 CPU with a promised 20 percent increase in performance on its predecessor, the Mali-G78 graphic processor and the Ethos-N8 NPU for on-device machine learning. The platforms are expected to power the next generation of 5G mobile devices and AI applications.
The new Arm Cortex-A78 CPU is designed to increase power efficiency over a smaller area. Arm said it brings a 20 percent increase in sustained performance over Cortex-A77-based devices within a 1-watt power budget, more efficient management of compute workloads and greater on-device machine learning performance.
The improved performance-per-watt also makes it well-suited for the emerging category of foldable devices with multiple and larger screens, the company said. Arm also introduced the Cortex-X1 with even better performance for the flagship smartphone and large-screen device category. This is the most powerful Cortex CPU to date and the first in its new Cortex-X Custom program for differentiating the traditional roadmap of Cortex products.
Next up in the graphic line-up is the Mali-G78. Built on the same Valhall architecture as last year's Mali-G77 GPU, the new delivers another 25 percent increase in graphics performance. The Mali-G8 can support up to 24 cores, thanks to asynchronous top level, tiler enhancements and improved fragment dependency tracking.
The Mali line also gets some differentiation, based on demand from customers. Alongside the premium Mali-G78, Arm is introducing a new sub-premium tier of GPUs. The first in this new tier is the Arm Mali-G68, which supports up to six cores with all the latest Mali-G78 features for a more affordable class of devices.
Finally, Arm presented the latest in its line of neural processing units, the Ethos-N78. This is designed to work a growing range of use cases relying on machine learning, from smartphones to smart home hubs. Arm said the N78 offers up to 25 percent more performance efficiency on its predecessor and configurations from 1 TOP/s up to 10 TOP/s.
