1. RVA23 compliance
2. RISC-V vector support with dual Vector Engine
3. Scalar and Vector Crypto
4. Multi-Cluster support
5. Hypervisor and AIA support
6. IOPMP and IOMMU

above are the highlights of AX66.


AX66 features a diverse and highly integrated design that includes security (with joint support for IOPMP and IOMMU), virtualization (Hypervisor), and vector acceleration. It delivers performance on par with the Arm Cortex-A76, making it an ideal application processor for edge AI devices such as smart surveillance cameras, edge analytics terminals in smart factories, or intelligent retail endpoints with voice/image processing capabilities.

Additionally, AX66 is well-suited to serve as the control core in AI infrastructure SoCs. For example, in a data center AI accelerator SoC, AX66 can act as the central processor responsible for managing the software stack, resource allocation, and communication protocols, coordinating operations between AI cores and memory subsystems.

Moreover, AX66 can be deployed alongside AX46MPV and AndesAiRE I370 to form a complete and flexible AI solution. In this configuration, AX46MPV focuses on high-throughput AI vector computation, I370 provides real-time inference and neural network acceleration, while AX66 serves as the central controller for application-level processing and system resource scheduling. The collaboration between these components greatly enhances overall intelligence and system efficiency. This architecture is especially well-suited for deployment in smart edge gateways, heterogeneous computing platforms, or AI SoCs requiring multicore heterogeneous processing.

In summary, with its flexible features and high-performance computing capabilities, AX66 can operate independently or serve as a key component in a comprehensive AI architecture, delivering significant value whether in edge devices or data center SoC platforms.