In order to meet the computing storage/computing storage requirements of data explosion and AIoT, Arm introduced Cortex-R82 real-time processor IP for a new generation of storage devices, and brought 64-bit architecture into Cortex-R series for the first time. With 40-bit memory addressing, it can be matched with 1TBDRAM, and can support larger SSD than the previous generation Cortex-R8.

▲ Computational storage is generated to cope with real-time applications such as data center, flight autopilot and image processing.

The so-called computational storage concept is aimed at the explosion of current data volume and the demand of AIoT for real-time data, such as flight autopilot, smart city, security monitoring and so on. By computing and AI identifying the data in the front end of the storage device, the valuable data content is filtered to reduce the amount of data sent to the cloud server by edge devices such as webcams, or the data can be analyzed and operated in real time by processing the data at the edge end.

▲ In order to ensure compatibility, Arm and SNIA cooperated to make Cortex-R82 architecture conform to the open computing storage specification.

Arm architecture accounts for 85% of the market share of hard disks and SSD controllers. In order to meet the needs of partners to integrate computing and storage in a single controller, it is planned to adopt a new generation Cortex-R82 architecture with high-performance 64-bit architecture, optional MMU/Memorymanagement to run Linux, and provide ML machine learning with NEON. Meet the architecture requirements of the computing storage controller, and work with SNIA's computing storage team to make Cortex-R82 conform to the open computing storage specification.

▲ Computing storage can reduce the delay of data transmission to the cloud for processing, which is more energy-saving and safer.

▲Cortex-R82 is Arm's first Cortex-R product with 64-bit architecture.

Compared with the current Cortex-R8, the overall performance of Cortex-R82 is doubled, and the ML performance with NEON is increased by 14 times, and it can support up to 8 cores of single processor. Due to the use of 40-bit memory addressing, Cortex-R8' s memory addressing ability is improved from 4GB to 1TB, which can meet the market demand in a few years.

▲ Cortex-R82 with MMU can be used to execute Linux.

▲Cortex-R82 can flexibly allocate the number of cores executing RTOS and Linux according to the load.

Cortex-R82 not only supports typical RTOS real-time operating system, but also can execute Linux system for operation processing and advanced machine learning functions as long as MMU is selected. At the same time, after matching MMU, Cortex-R82 can flexibly allocate cores for data stream processing or operation. For example, most cores are used for data aggregation of RTOS when data is pouring in, and resources are allocated to Linux for operation and analysis when data input is less.