ARM’s new Cortex-A75, A55 processors along with the Mali-G72 GPU have been announced ahead of Taiwan’s Computex event, where PC makers showcase major launches, products. AR essentially provides the architecture and design for processors, that is then used by Qualcomm, Samsung, Apple for their own mobile processor chipsets.
With the new processor design, ARM is also focusing on machine learning and artificial intelligence, and says these processors will be ready to carry out these kind of tasks.
The ARM Cortex-A75 processor is an improvement over the earlier Cortex-A72, A73 processor with improved performance, efficiency, and also improvements for machine learning, says the company. It can be used in smartphones, “intelligent home devices like DTV, servers, and automotive applications,” claims the company.
ARM is claiming the Cortex-A75 processor offers 50 per cent improvement in performance and multi-core capabilities compared to the earlier version. In case of the Cortex-A55, ARM says this is their most efficient processor in the mid-range series, and offers 2.5x improved performance efficiency compared to the current Cortex-A53 based devices.
It also announced the new Mali-G72 GPU, which offers up to 40 boost in performance compared to the previous series, and is ready for mobile virtual reality experiences as well. It is also optimised to provide an efficient performance for Machine Learning, says ARM.
On the artificial intelligence and machine learning side of things, ARM’s new Cortex-A75, Cortex-A55 processor is built on the company’s DynamIQ technology, and the CPU can take advanced workloads including machine learning as well.
According to ARM’s announcement, the idea with this focus on Artificial intelligence is to move away from the cloud-centric approach for AI to more a real-time on device experience, which will also ensure better privacy and security around AI.
ARM’s Cortex-A75 and Cortex-A55 comes with detailed instructions for AI performance tasks with this new DynamIQ technology. ARM says it wants to boost AI performance by 50x over the next three to five years.